Background

Decisional conflict, a pivotal concept in healthcare decision-making, refers to the psychological distress experienced when individuals face choices involving inherent risks, uncertain outcomes, and potential conflicts with personal values1. It is characterized by uncertainty about options, inadequate information, ambiguity in personal values, and perceived external pressures2. Elevated decisional conflict has been linked to delayed choices, post-decision regret, and strained clinician–family relationships, all of which can compromise care quality3. In critical care, where life-altering decisions must be made rapidly, decisional conflict may be particularly pronounced4.

Extracorporeal membrane oxygenation (ECMO) is an advanced life-support modality that provides temporary respiratory and/or circulatory support for patients with severe cardiopulmonary failure via a membrane oxygenator5. Once experimental, ECMO is now an established rescue therapy for refractory respiratory and cardiac failure. Utilization surged during the COVID-19 pandemic, with more than 10,000 patients worldwide supported between 2020 and 20226. Nonetheless, ECMO presents major challenges: initiation requires precise timing within a narrow therapeutic window, and complications such as bleeding, thrombosis, and infection remain common7,8. In cardiogenic shock, pooled bleeding and renal failure rates reach ~ 49% and 51%, respectively9. Financial burdens are also considerable: U.S. hospital costs typically range from US$50,000 to US$350,000 per case, depending on indication and duration, with daily costs averaging about US$600010. These medical and economic complexities highlight the multidimensional nature of ECMO decision-making.

In practice, family members frequently serve as surrogate decision-makers, navigating emotionally charged decisions that balance survival probabilities, quality of life, financial feasibility, and ethical considerations under severe time constraints11. Qualitative studies reveal that families often experience information overload, emotional distress, and value conflicts when confronted with ECMO decisions12, conditions that may exacerbate decisional conflict, delay initiation, or complicate clinician–family communication13. Ethical issues, such as informed consent, resource allocation, and continuation or withdrawal of support, further complicate the process14,15. Many surrogates struggle to comprehend long-term implications, risks of complications, and potential withdrawal scenarios16.

Despite their central role, few studies have examined decisional conflict among family members during ECMO initiation, particularly in non-Western contexts where cultural norms and family dynamics strongly shape decision-making17,18. Existing research is limited, often qualitative and small-scale, leaving the prevalence and associated factors of decisional conflict poorly understood.

Informed consent is a fundamental ethical and legal principle in modern medicine. For patients undergoing high-risk interventions such as ECMO, the right to individual decision-making must be respected. Whenever feasible, patients should be fully informed about the risks, benefits, and alternatives to ECMO therapy. However, in critical care contexts, particularly during acute deterioration or cardiopulmonary arrest, many patients are incapacitated and unable to express their preferences. In such cases, surrogate decision-makers—typically family members—are engaged to represent the patient’s presumed wishes, values, and advance directives17. This surrogate involvement highlights the ethical complexity surrounding ECMO initiation and underscores the importance of comprehensive communication strategies by healthcare professionals to ensure decisions align with the patient’s best interests and values.

Therefore, this study aimed to quantify the prevalence of decisional conflict among Chinese family caregivers prior to ECMO initiation and to identify associated biopsychosocial factors using validated instruments. By addressing this gap, the study provides quantitative evidence to guide context-specific decision support and strengthen physician–family communication in critical care.

Methods

Design and participants

This cross-sectional study employed a convenience sampling method to recruit 169 family members of patients for whom ECMO initiation was deemed necessary by the treating physicians. Participants were enrolled from the Department of Critical Care Medicine at a tertiary hospital in Jiangxi Province, China, between January 2024 and April 2025. A participant flow diagram, developed in accordance with the STROBE guidelines, is presented in Fig. 1. Sample size was determined using the “10 events per variable (EPV)” criterion, which is a widely accepted rule of thumb for logistic regression analysis. It recommends that at least 10 outcome events be included for each predictor variable to ensure model stability and reduce the risk of overfitting19. Considering seven candidate predictors in the regression model, an anticipated 52% prevalence of decisional conflict (derived from pilot data), and a 15% attrition rate, the minimum required sample size was calculated to be 169. In this study, ECMO was initiated in patients who met established clinical indications. Specifically, veno-venous (VV) ECMO was applied for patients with severe, refractory respiratory failure (e.g., acute respiratory distress syndrome unresponsive to conventional mechanical ventilation), while veno-arterial (VA) ECMO was used for patients with cardiogenic shock or refractory cardiac arrest. Among the patients whose family members were surveyed, 139 received VV ECMO and 30 received VA ECMO. These groups represent distinct clinical contexts, as VV ECMO primarily addresses isolated respiratory failure, whereas VA ECMO is typically associated with multi-organ dysfunction and higher illness severity.

Eligibility criteria

Inclusion criteria for family members were

  1. (1)

    Patients met ECMO indications and had signed informed consent;

  2. (2)

    Direct relatives of the patient (decision-making hierarchy: spouse > parents or children > siblings);

  3. (3)

    Age ≥ 18 years, with adequate communication and comprehension abilities;

  4. (4)

    Full participation in physician–family communication prior to ECMO initiation;

  5. (5)

    Provision of written informed consent for participation in this study.

Exclusion criteria were

  1. (1)

    Ongoing involvement in medical disputes or legal proceedings;

  2. (2)

    Previously diagnosed cognitive dysfunction or mental illness;

  3. (3)

    Severe hearing or language impairment.

Ethical considerations

The study protocol was approved by the hospital’s ethics committee (approval number: 1IT[2024] Clinical Ethics Review No. 339), and all procedures were conducted in accordance with relevant guidelines and regulations. Written informed consent was obtained from all participants.

Clinical decision-making

In this study, the initiation of ECMO was a clinical decision made by the treating physicians after a comprehensive risk–benefit assessment. Evaluation of the patient’s presumed wishes and values was sought to support this process; when such evaluation was not feasible, surrogate decision makers were consulted. Ultimately, the indication to initiate ECMO was determined by clinicians.

Data integrity

Cases with missing values were handled using listwise deletion.

Measurements

General information

Demographic and clinical information were collected using a self-designed questionnaire developed through a literature review. Variables included caregiver gender, age, education level, marital status, monthly per capita family income, place of residence, level of understanding of the patient’s condition and ECMO treatment, and prior ICU caregiving experience. This section was self-reported by family members.

Decisional conflict scale

Decisional conflict was assessed using the 16-item DCS developed by O’Connor20 and validated in Chinese by Li Yu21. It includes five dimensions: being informed, clarifying values, support, uncertainty, and effective decision-making. Items are scored on a 5-point Likert scale (0 = “yes” to 4 = “no”), with higher scores indicating greater conflict. Total scores are converted to a 0–100 scale, with ≥ 25 indicating decisional conflict. The Chinese version has demonstrated high internal consistency (Cronbach’s α > 0.90) and construct validity. In this study, Cronbach’s α = 0.922.

Decision fatigue scale

Decision fatigue was measured using the 9-item DFS developed by Hickman et al.22 and translated and validated in Chinese by Pan Guocui et al.23 for ICU family caregivers. Responses are rated on a 4-point Likert scale (0 = “strongly disagree” to 3 = “strongly agree”), with total scores ranging from 0 to 27. Higher scores reflect greater fatigue. The Chinese version has demonstrated acceptable reliability and convergent validity. In this study, Cronbach’s α = 0.912.

Decision preparedness scale

Preparedness for decision-making was assessed with the 10-item DPS, originally developed by O’Connor and revised by Bennett24. The Chinese version, validated by Wan Junli et al.25, has shown content validity and criterion-related validity. Items are rated on a 5-point Likert scale (1 = “none at all” to 5 = “very much”). After conversion to percentages, scores < 60 indicate poor preparedness. In this study, Cronbach’s α = 0.959.

Perceived social support scale

Perceived social support was assessed with the 12-item PSSS developed by Zimet et al.26 and translated and validated by Jiang Qianjin et al.27. The Chinese version retains two dimensions: family internal support (4 items) and external support (8 items). Items are rated on a 7-point Likert scale, with total scores ranging from 12 to 84. Higher scores reflect stronger support. The Chinese version has demonstrated factorial validity and cross-cultural applicability. In this study, Cronbach’s α = 0.902.

Wake forest physician trust scale

Trust in physicians was measured using the WFPTS developed by Hall et al.28 and validated in Chinese by Dong Enhong et al.29. The scale includes 10 items across two dimensions (benevolence and technical competence), rated on a 5-point Likert scale. Higher scores indicate greater trust. Prior validation studies of the Chinese version have demonstrated satisfactory construct validity and reliability. In this study, Cronbach’s α = 0.812. higher scores indicate lower trust. All scales used validated Chinese versions; scoring followed manuals.

Data collection and quality control methods

Participant screening was conducted in two stages. First, researchers reviewed patient medical records to identify eligible cases. Second, family surrogate decision-makers were contacted for further eligibility confirmation. Eligible family members were invited to a private consultation room before ECMO initiation, where the study purpose and procedures were explained and written informed consent obtained.

Trained researchers provided standardized instructions for questionnaire completion, answered questions on site, and checked each questionnaire immediately after completion. Cases with missing responses were excluded listwise.

In the Chinese cultural context, surrogate decision-making is often shared among multiple family members. To ensure consistency, the questionnaire was completed by the primary decision-maker, identified according to the family decision-making hierarchy (spouse > parents/children > siblings). This was acknowledged as a potential source of respondent bias.

Data analysis

Statistical analyses were performed using Microsoft Excel 2021 and IBM SPSS Statistics (version 26.0). Continuous variables with a normal distribution were presented as mean ± standard deviation (SD), while non-normally distributed data were expressed as median (interquartile range, IQR). Categorical variables were summarized using frequencies and percentages. Based on Decisional Conflict Scale (DCS) scores, family members were categorized into two groups: “No decisional conflict” (< 25 points) and “Decisional conflict present” (≥ 25 points). Univariate analyses employed Mann-Whitney U tests for non-parametric data, independent samples t-tests for normally distributed continuous variables, and Pearson’s chi-square tests for categorical variables. Variables demonstrating statistical significance (P < 0.05) in univariate analyses were subsequently entered into binary logistic regression modeling. The statistical significance threshold was set at α = 0.05 (two-tailed).

Results

General information of patients’ families

The demographic characteristics of the 169 family caregivers are summarized in Table 1. The majority were female, middle-aged, and had at least a high school education. Most reported household monthly incomes between RMB 5000 and 10,000, and 86.3% had at least a basic understanding of ECMO. Only 21.9% reported prior ICU caregiving experience.

Descriptive statistics of key scales

The mean Decision Preparedness Scale score was 28.64 ± 6.66, indicating a moderate level of preparedness. The mean Decision Fatigue Scale score was 16.39 ± 3.37, reflecting a moderately high level of fatigue. The mean Decisional Conflict Scale score was 29.63 ± 6.73, indicating high decisional conflict. The mean Physician Trust Scale score was 28.35 ± 7.36, suggesting an intermediate level of trust, and the mean Social Support Scale score was 49.76 ± 12.87, reflecting moderately high social support.

Univariate analysis

Univariate analyses showed that decisional conflict was significantly associated with monthly household income, level of ECMO knowledge, decision fatigue, decision preparedness, physician trust, and social support (all P < 0.05). In contrast, education level, prior ICU experience, and payment method were not significantly associated with conflict (Table 2). For example, higher decision fatigue scores and lower preparedness scores were associated with higher decisional conflict. Similarly, lower physician trust and weaker social support were associated with greater conflict.

Multifactorial analysis

Variables significant in the univariate analysis were entered into a binary logistic regression model, with the presence of decisional conflict (no = 0, yes = 1) as the dependent variable. The model was significant (Omnibus χ² test, P < 0.05) and explained 64.5% of the variance in decisional conflict (Table 3). Results indicated that: Higher decision fatigue was associated with greater odds of decisional conflict (OR = 1.44, 95% CI 1.00–2.06). Higher decision preparedness was associated with lower odds of conflict (OR = 0.85, 95% CI 0.72–0.99). Lower physician trust (note: higher WFPTS scores indicate lower trust) was associated with greater conflict (OR = 3.15, 95% CI 1.51–6.57). Lower social support was associated with greater conflict (OR = 0.44, 95% CI 0.28–0.70). Lower ECMO knowledge was associated with greater conflict (OR = 0.18, 95% CI 0.06–0.51). These findings suggest that decisional conflict among ECMO caregivers is strongly associated with both cognitive-emotional factors (fatigue, preparedness) and relational/contextual factors (trust, social support, knowledge). Detailed coding procedures for categorical variables are provided in Supplementary Table 1.

Table 1 Demographic characteristics of family caregivers (n = 169)
Full size table
Table 2 Factors associated with decisional conflict among ECMO caregivers (n = 169)
Full size table
Fig. 1
figure 1

STROBE diagram of participant flow.

Full size image
Table 3 Multivariate analysis of factors influencing decision conflict among patients’ family members
Full size table

Discussion

This study found that 63.3% of family members reported decisional conflict before ECMO initiation, with a mean score of 29.63 ± 6.73. This prevalence is higher than previously reported among general ICU family members (24.32 ± 15.19)30 and medical ICU family members (21.9 ± 14.8)31. The elevated conflict likely reflects the complexity of ECMO, a costly and high-risk therapy with uncertain prognosis and strict eligibility criteria, which creates substantial information asymmetry between physicians and families. In China, advance care planning is rarely practiced—only 12.3% of end-stage patients participate in such discussions—leaving families without prior guidance during urgent decision-making. Moreover, the need to consider multi-organ support, particularly in VA ECMO cases, may intensify decisional burden under time pressure32.

Lower levels of understanding of the patient’s condition and ECMO treatment were associated with higher decisional conflict, consistent with prior research32,33. Insufficient comprehension of ECMO’s potential benefits, risks, complications, and prognosis can heighten uncertainty and anxiety34. Evidence indicates that comprehensive physician–family communication improves treatment understanding and reduces decisional conflict35. Transparent communication also helps to foster trust, which is an essential factor in reducing decisional burden (Table 4).

Table 4 Variable coding scheme for independent variables.
Full size table

Decision fatigue was positively associated with decisional conflict, suggesting that higher levels of fatigue correspond to greater conflict. Decision fatigue, defined as reduced cognitive capacity and increased emotional burden from repeated decision-making32, may be exacerbated in ECMO contexts by information overload and prognostic uncertainty. Interventions such as simplifying information delivery, pacing discussions, offering structured emotional support, and distributing decision roles among family members may help alleviate fatigue36.

Lower preparedness for decision-making was also associated with higher decisional conflict, consistent with prior ICU studies28. Many families are suddenly thrust into surrogate roles with little time to prepare. Decision aids, structured risk–benefit explanations, and scheduled family meetings may improve preparedness and reduce conflict.

Higher social support was associated with lower decisional conflict, in line with Lee et al.37. Support networks provide emotional comfort, practical help, and informational resources, reducing psychological stress38. Strong internal support facilitates consensus-building, while external support contributes knowledge and encouragement. Hospitals may consider strengthening multi-level support systems, including family counseling services and peer-support programs.

Physician trust was also associated with decisional conflict. Importantly, in this study, higher WFPTS scores indicated lower levels of trust. Therefore, the observed OR > 1 suggests that lower trust was associated with greater conflict. Trust plays a pivotal role in facilitating complex treatment decisions39. Building trust requires not only accurate and timely information but also the demonstration of professional competence and empathy by clinicians.

Interestingly, higher monthly household income was also associated with decisional conflict. This may be explained by wealthier families being more informed, having higher expectations for treatment success, and being more sensitive to the potential financial implications of ECMO therapy.

These findings are consonant with Rajsic and colleagues’ recent ethical analyses, which argue that decisions about ECMO—especially extracorporeal cardiopulmonary resuscitation (eCPR)—should weigh short-term survival gains against realistic neurological prognosis, anticipated quality of life, and the patient’s known or presumed preferences, rather than survival alone40. In the time-pressured context of eCPR—where prospective informed consent is rarely feasible—the authors emphasize structured engagement with surrogate decision-makers, early and transparent disclosure of expected neurological outcomes and potential withdrawal thresholds, and clear documentation of advance directives to uphold autonomy and reduce decisional burden41. Beyond individual patients, they highlight distributive-justice and societal considerations: as eCPR is integrated into Advanced Life Support pathways, systems may confront an increase in survivors with severe disability and long-term care needs, shifting burdens to families and healthcare resources; ethically sound adoption therefore requires transparent candidacy criteria, data-informed prognostication, and access to institutional ethics support42. Finally, in cases of unfavorable neurological outcome after eCPR, Rajsic et al. underscore the ethical complexity of subsequent organ-donation discussions and call for strict separation of treatment-withdrawal decisions from donation considerations, robust consent processes, and safeguards against conflicts of interest to preserve the dead-donor rule and public trust43.

This study provides quantitative evidence on decisional conflict among Chinese ECMO family caregivers, filling a gap in a literature base that has been dominated by qualitative work.

Limitations

Several limitations should be acknowledged. First, this study was conducted in a single tertiary hospital in Jiangxi Province using a convenience sampling strategy, which limits the generalizability of the findings to other regions, healthcare systems, or cultural contexts. ICU organization and psychosocial resources may vary substantially across settings. Second, although both VV and VA ECMO patients were included, these represent clinically distinct populations with different illness severity and decision-making challenges, which may have influenced the findings. Third, surrogate decision-making in China is often collective, yet our survey was typically completed by a primary family member, which may have introduced respondent bias. Finally, the cross-sectional design precludes causal inference. Future multicenter and longitudinal studies are warranted to validate these results and to explore targeted interventions to reduce decisional conflict.

Conclusions

A substantial proportion of family members of ECMO patients experience decisional conflict prior to treatment initiation. While ECMO initiation is ultimately a clinical decision made by physicians, family caregivers’ understanding of the patient’s condition and treatment, their preparedness, levels of fatigue, trust in physicians, and access to social support are closely associated with how they engage in the decision process. These findings underscore the professional obligation to provide complete and comprehensible information, and they highlight the value of structured communication strategies (e.g., scheduled family meetings, standardized ECMO information sheets), decision-support tools, and psychosocial interventions. Such measures may help reduce decisional conflict, support alignment with the patient’s presumed wishes, and improve the efficiency of decision-making in ICU contexts. Future multicenter and longitudinal studies are warranted to validate these findings and explore culturally tailored interventions.