CARE-ATLANTIC: understanding cross cultural differences in goals of care; a transatlantic study in critical care medicine

Abstract

Background

Medical paternalism, where physicians decide based on perceived patient interests, contrasts with the value of patient autonomy, which emphasizes patients’ rights to make informed choices about their medical care. This study explores how trainees from cultures emphasizing these contrasting values interpret limitations on life-sustaining treatments, focusing on potential differences in decision-making.

Methods

A survey was designed with two clinical vignettes and trainees rated their agreement or disagreement with 21 different treatment choices on a 5-point Likert scale. Surveys were distributed via Fellowship Program Directors in the U.S. and the ANJMIR in France. Statistical analysis used the Mann–Whitney U test to compare responses across countries and code status groups.

Results

A total of 217 trainees responded. Both American and French critical care trainees made different decisions based on patient code status alone, with the Americans being very polarized in their decisions and the French less so. Significant variations across interventions reflected cultural influences on medical decision-making.

Conclusions

This study reveals significant cultural differences in decision-making between French and American critical care trainees. Americans align more closely with patient preferences, while French trainees show greater variability, highlighting the impact of cultural norms on medical practices and patient care.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12910-026-01554-9.

Introduction

Paternalism in medicine is the practice wherein the physician makes decisions based on what he or she believes to be in the patient’s best interests, even for those patients who are capable of making their own decisions [1–3]. This practice contrasts with the concept of patient autonomy which holds that it is the right of the patient to make informed decisions about their medical care without undue influence from their healthcare provider. The focus on patient autonomy has become increasingly prominent over the last half century [4, 5] particularly in the United States, where the legacy of the Belmont Report [6] and the broader bioethics movement established autonomy as the dominant principle of medical decision-making, intended to safeguard against physician authority and guarantee patient self-determination. In France, by contrast, clinical practice remains more deeply rooted in the Hippocratic ethos of beneficence. Until the early twenty-first century, French physicians functioned as custodians of patient welfare, often reinterpreting or overriding patient preferences deemed medically inappropriate. The enactment of the Kouchner Law on March 4, 2002, established patient autonomy as the guiding decision-making model within the French healthcare system.

These practice differences are reflected in the ethical codes each country follows. The code of ethics in France states that, “when the patient is unable to express their wishes, the attending physician is required to respect the will expressed by the patient in advance directives, except […] if the attending physician considers the advance directives to be clearly inappropriate or not in accordance with the medical situation.” [7] While this directive is an example of the paternalism that the medical world has tried to move away from, it is based on the sentiment that the patient is vulnerable and there is an underlying assumption of beneficence (non-maleficence) from the medical team. It reflects the perspective that allowing medical decisions to be driven mainly by the patient and/or their family may deprive that same patient and family of the expert and professional advice that they need and deserve to make medical decisions [8].

In contrast, the ideas outlined in the Code of Medical Ethics issued by the American Medical Association place the patient and their surrogate more at the center of medical decisions and emphasize patient autonomy. For example, the American code stipulates that, “the surrogate should make decisions to withhold or withdraw life-sustaining interventions when the patient lacks decision-making capacity,” demonstrating that physicians are not intended to be the main drivers in these decisions [9].

Accordingly, studies have found that paternalism is still a significant driver in the enactment of end-of-life decisions in Europe [10]. But it is notable that even with increasing attention being paid to implementing orders to limit life sustaining treatment in the United States [11], the interpretation of such limitations varies widely among physician trainees in the United States [12–14]. And, Intensive Care Units (ICUs) in the United States vary significantly in the management of patients admitted with treatment limitations in their advance directives [15]. Even when family is involved and respect for autonomy is espoused, research has shown that American clinicians often do not make recommendations based on patients’ values and preferences [16].

Multiple studies have documented different perspectives around advance directives and decisions to forgo life-sustaining treatment [17–19] between France and the United States^.. The purpose of our CARE-ATLANTIC study was to evaluate how French and American physicians training to become critical care specialists interpret the scope and meaning of Do Not Resuscitate/Do Not Intubate (DNR/DNI) orders, as this issue is particularly pertinent to this physician population. We hypothesized that when presented with a hypothetical patient, American critical care trainees would interpret a DNR/DNI code status to signify that a patient desired less aggressive care overall, and that French critical care trainees would be more likely to make decisions that are directly in conflict with a patient’s desired code status.

Methods

CARE-ATLANTIC survey development

To conduct this research, we designed a survey instrument that presented two clinical vignettes involving patients who required hospitalization. Each vignette described a patient admitted to the regular hospital ward who then decompensated and was transferred to the Intensive Care Unit (ICU). Critical care trainees were asked about their level of agreement with 21 treatment decisions for each patient. To reduce survey error, we performed cognitive interviewing [20] with critical care trainees in both the United States and France and used their input for instrument finalization.

We offered a variety of treatment decisions that ranged from noninvasive (intravenous fluid administration, blood and urine cultures, goals of care discussion) to therapies that may be perceived as moderately aggressive (central venous catheter placement, esophogastroduodenoscopy, dialysis, surgery consultation), to interventions that may be perceived as very aggressive (cardiopulmonary resuscitation and emergent intubation). The majority of the treatment decisions were common to both vignettes, and several were offered in both the general hospital ward and ICU settings. Respondents used a 5-point Likert scale (strongly agree to strongly disagree) to indicate whether they would choose to provide each intervention if they were caring for the patient in the vignette. We created two versions of the survey (A and B). Each version included the same two clinical vignettes, differing only in the code status (Full Code or DNR/DNI) assigned to the theoretical patient. Each survey version included one patient vignette with Full Code status and one with DNR/DNI status. We also collected basic demographic data about respondents and their training programs.

Study population and survey distribution

Study participants were medical trainees who had chosen a critical care pathway. In the United States, the trainees had already had a minimum of three years of training in either Internal Medicine, General Surgery, Emergency Medicine, Anesthesiology, or Neurology residencies, and were pursuing additional training in Critical Care Fellowship.

In France, participants had already completed two years of medical school coursework, followed by four years of combined learning consisting of university course work and clinical internships. At the time that they received the survey, French trainees were enrolled in a residency lasting between 1 and 5 years, during which time they practice medicine full-time in hospitals under the supervision of senior certified physicians.

In the United States, surveys were sent to Fellowship Program Directors who then distributed them to their trainees. A reminder email was sent two weeks after the first email and then a final one two weeks later. In France, surveys were sent via email, directly to the trainees via the national association of young intensive care residents, the Association des Jeunes Médecins Intensivistes et Réanimateurs (ANJMIR). Reminder emails were sent at the same time intervals as with the US trainees.

Survey administration

For each distribution method, the questionnaire was accessible via a link that directed the trainee to vignettes number 1 (Ms. L) and number 2 (Ms. Y).

To provide randomization in France, we collected the last names of each eligible participant, alphabetized them, and split the list in half. The first half received survey A, where Ms. L has a Full Code status and Ms. Y is DNR/DNI, and the second half received survey B, wherein Ms. L is DNR/DNI and Ms. Y is Full Code.

To ensure randomization among United States participants, we compiled a list of Program Directors for the various critical care fellowships (Pulmonary/Critical Care Medicine, Critical Care Medicine, Surgical Critical Care, Anesthesia Critical Care and Neuro Critical Care) split it, and shared survey A with half and survey B with the other half, and asked them to distribute their respective surveys to their trainees.

We received responses from trainee doctors across both countries, with at least 10 participants from each region in each country, except for two regions in the United States: Puerto Rico, with 1 participant, and the Southwest, with 4 participants.

Ethics approval and consent to participate

This project was validated by the institutional review board of Beth Israel Deaconess Medical Center (2023P000755) and the university hospital center of Montpellier (IRB 202301458) and consent to participate was waived. This project was performed in accordance with the Declaration of Helsinki.

Statistical analysis

Data analysis was conducted using the Mann–Whitney U test to compare engagement levels between the American and French respondents within the Full Code and DNR/DNI categories, and vice versa. The Mann–Whitney U test was selected because of the ordinal nature and non-normal distribution of the response data. The Holm method was used to adjust p-values for multiple comparisons, as it is less conservative than the Bonferroni adjustment and reduces the risk of type II errors [21, 22]. The significance threshold was set at 0.05, with adjusted p-values below this threshold considered statistically significant. All analyses were performed using R Statistical Software (v4.2.2; R Core Team 2022).

Results

A total of 217 trainees responded to the survey—123 in France (79 completed) and 94 in the U.S. (78 completed). The majority of U.S. respondents were aged between 29 and 35 years old and were more likely to be men and in their 5th or 6th year of training post medical school. French respondents were aged between 25 and 28, with an equal gender distribution and were less advanced in their training (Table 1).

Table 1.

Characteristics of respondents

US (n = 94)
Age	24 and under	1
	25 to 28	0
	29 to 34	59
	35 and over	30
	Without answer	4
Gender	Female	38
	Male	52
	Neither	0
	Wishes not to reply	0
	Other	0
	Without answer	4
Geography	Northeast	33
	Southeast	21
	Southwest	4
	Midwest	21
	West	10
	Puerto Rico	1
	Without answer	4
Background	PGY-3	1
	PGY-4	18
	PGY-5	28
	PGY-6	28
	PGY-7	6
	PGY-8	6
	PGY-9	2
	PGY-10	0
	Other	1
	Without answer	4
Completed fellowship training	Yes	11
	No	79
	Without answer	4
Worked prior as an attending physician	Yes	21
	No	69
	Without answer	4
Year of fellowship	First year	52
	Second year	23
	Third year	12
	Fourth year	1
	Other	2
	Without answer	4
France (n = 123)
Age	24 and under	14
	25 to 28	75
	29 to 34	29
	35 and over	2
	Without answer	3
Gender	Female	60
	Male	61
	Wishes not to reply	0
	Other	0
	Without answer	2
Year of study	1 st year of residency	19
	2nd year of residency	33
	3rd year of residency	24
	4th year of residency	22
	5th year of residency	22
	Other	3
Geography	South (Auvergne-Rhone-Alpes, PACA, Occitanie)	63
	West (Bretagne, Centre-Val de Loire et Pays-de-la-Loire, Normandie, Nouvelle Aquitaine)	24
	North & east (Bourgogne-Franche-Comté, Grand Est, Hauts de France)	23
	Paris & Martinique-Guadeloupe, Indian Ocean	11
	Without answer	2

The effect of patient elected code status on the provision of cardiopulmonary resuscitation (Figs. 1 and 2)

Fig. 1.

Decision on intubation and CPR: proportion of the different responses to all relevant questions throughout the progression of vignette 1. Polarized and non-polarized response proportions across vignette 1 progression

Fig. 2.

Decision on intubation and CPR: proportion of the different responses to all relevant questions throughout the progression of vignette 2. Polarized and non-polarized response proportions across vignette progression

Vignette #1 describes Ms. L, a 70-year-old woman with Clostridium difficile colitis. Critical care trainees from both countries were likely to withhold CPR and intubation with DNR/DNI status. However, French responses were less polarized than Americans'. French trainees were less likely to strongly agree with intubating with Full Code status or strongly disagree with withholding therapy with DNR/DNI status. They were also less likely to strongly agree with CPR with Full Code status or strongly disagree with withholding it with DNR/DNI status (Table 2 and Supplemental Table 2).

Table 2.

Responses to vignette n°1 (comparison of American and French responses)

Ms. L is a 70 year old woman with hyperlipidemia who recently received a course of oral clindamycin for a dental infection. She presents to the emergency room with profuse diarrhea and is diagnosed with clostridium difficile colitis and acute kidney injury. She is started on appropriate antibiotics. On admission Ms. L elects a code status of FULL CODE or DNR/DNI. Ms. L is initially admitted to the regular hospital ward
Question	Engagement	Full code USA	Full code France	P-value Full Code	adjusted P-value	DNI/ DNR USA	DNI/ DNR France	P-value DNI/ DNR	adjusted P-value
1. If Ms. L develops renal failure and becomes anuric and acidotic, I would initiate plans for dialysis	Strongly agree	30	35	0.583	1	13	11	0.68	1
	Somewhat agree	11	11			23	17
	Neither	2	1			4	13
	Somewhat disagree	0	0			2	4
	Strongly disagree	0	0			0	0
2. I would discuss goals of care with Ms. L	Strongly agree	33	38	0.638	1	39	40	0.492	1
	Somewhat agree	6	5			3	3
	Neither	3	4			0	2
	Somewhat disagree	1	0			0	0
	Strongly disagree	0	0			0	0
3. If Ms. L develops septic shock secondary to clostridium difficile colitis, I would consult Surgery for consideration of a colectomy for source control	Strongly agree	34	20	0.001	0.023	19	13	0.215	1
	Somewhat agree	6	19			15	21
	Neither	2	2			4	6
	Somewhat disagree	0	5			4	3
	Strongly disagree	1	1			0	1
	No answer	0	0			0	1
4. If Ms L develops a severe hypoxemic respiratory failure that does not respond adequately to medical therapy, I would recommend intubation	Strongly agree	34	28	0.04	1	2	4	0.027	0.73
	Somewhat agree	8	15			2	9
	Neither	1	3			5	10
	Somewhat disagree	0	1			14	7
	Strongly disagree	0	0			19	15
5. If Ms. L has a respiratory arrest, I would intubate her	Strongly agree	41	29	0	0.005	0	0	0.001	0.029
	Somewhat agree	2	16			0	6
	Neither	0	2			2	3
	Somewhat disagree	0	0			4	12
	Strongly disagree	0	0			36	24
6. If Ms. L is found pulseless, I would perform CPR	Strongly agree	41	32	0.001	0.029	0	0	0	0
	Somewhat agree	2	12			0	5
	Neither	0	3			0	4
	Somewhat disagree	0	0			0	12
	Strongly disagree	0	0			42	24
Ms. L is transferred to the floor of your hospital where she can receive intermittent hemodialysis. She becomes anuric and acidotic and develops hypotension when intermittent hemodialysis is attempted, so is transferred to the Intensive Care Unit for Continuous Renal Replacement Therapy. She is on day 3 of her ICU stay
7. If Ms. L develops hypotension unresponsive to fluid resuscitation I would start vasopressors	Strongly agree	40	37	0.044	1	30	25	0.262	1
	Somewhat agree	1	9			9	14
	Neither	0	0			2	0
	Somewhat disagree	1	0			0	1
	Strongly disagree	0	0			0	1
8. If Ms. L develops septic shock secondary to C. difficile colitis, I would consult Surgery for consideration of a colectomy for source control	Strongly agree	35	22	0.001	0.023	18	15	0.215	1
	Somewhat agree	3	16			15	21
	Neither	2	5			5	3
	Somewhat disagree	1	3			3	2
	Strongly disagree	1	0			0	0
9. If Ms. L has a respiratory arrest, I would intubate her	Strongly agree	40	29	0	0.005	0	0	0.001	0.029
	Somewhat agree	2	14			0	2
	Neither	0	3			1	2
	Somewhat disagree	0	0			5	16
	Strongly disagree	0	0			35	21
10. If Ms. L becomes pulseless, I would perform CPR	Strongly agree	40	30	0	0.016	0	0	0	0
	Somewhat agree	2	12			0	2
	Neither	0	4			0	3
	Somewhat disagree	0	0			1	13
	Strongly disagree	0	0			40	23

Vignette #2 describes Ms. Y, a 38-year-old bedridden multiple sclerosis patient with a wound infection. Both groups withheld CPR and intubation in DNR/DNI status and provided these with full code status. French trainees showed less polarization, with 40% disagreeing with intubation and 30% disagreeing with CPR despite Full Code status. Americans were more likely to strongly agree with intubation for hypoxemia and respiratory arrest, and CPR for pulselessness with Full Code status. With DNR/DNI patients, French trainees were less likely to strongly disagree with intubation for respiratory arrest and CPR for pulselessness (Table 3 and Supplemental Table 3).

Table 3.

Responses to vignette n°2 (comparison of American and French responses)

Ms. Y is a 38 year-old woman with multiple sclerosis who has been bedridden for 3 years, and has extensive sacral pressure ulcers with chronic osteitis. She presents to the emergency room with sepsis in the setting of a new wound infection. She is initiated on broad spectrum antibiotics. On admission, she elects a code status of FULL CODE or DNR/DNI. You first evaluate Ms. Y right after her admission to the medicine ward
Question	Engagement	Full code USA	Full code France	P-value Full Code	adjusted P-value	DNI/ DNR USA	DNI/ DNR France	P-value DNI/ DNR	adjusted P-value
1. I would discuss goals of care with Ms. Y and her healthcare proxy	Strongly agree	39	38	0.32	1	39	39	0.188	1
	Somewhat agree	1	3			1	2
	Neither	0	0			0	0
	Somewhat disagree	0	0			0	1
	Strongly disagree	0	0			0	0
2. I would consult surgery for management of Ms. Y’s wound infection	Strongly agree	28	34	0.184	1	27	23	0.298	1
	Somewhat agree	11	6			9	15
	Neither	0	0			4	3
	Somewhat disagree	1	1			0	1
	Strongly disagree	0	0			0	0
3. I would transfer Ms. Y to the Intensive Care Unit if she remains hypotensive after fluid resuscitation	Strongly agree	36	18	0	0	33	7	0	0
	Somewhat agree	4	16			6	16
	Neither	0	4			1	8
	Somewhat disagree	0	3			0	8
	Strongly disagree	0	0			0	3
4. If Ms Y develops a severe hypoxemic respiratory failure that does not respond adequately to medical therapy, I would recommend intubation	Strongly agree	24	6	0	0.001	2	1	0.577	1
	Somewhat agree	10	12			2	2
	Neither	1	10			5	4
	Somewhat disagree	2	9			9	17
	Strongly disagree	3	4			22	18
5. I would intubate Ms. Y if she had a respiratory arrest	Strongly agree	34	6	0	0	1	1	0.002	0.041
	Somewhat agree	2	12			0	0
	Neither	2	7			0	2
	Somewhat disagree	1	9			5	17
	Strongly disagree	1	7			34	22
6. I would perform cardiopulmonary resuscitation on Ms. Y if she became pulseless	Strongly agree	31	8	0	0	0	1	0	0.003
	Somewhat agree	6	8			0	0
	Neither	1	13			0	2
	Somewhat disagree	1	7			2	14
	Strongly disagree	1	5			38	25
Ms. Y is transferred to the ICU with septic shock due to her wound infection. She is being treated with broad spectrum antibiotics, insulin, vasopressors and oxygen. She is currently on day 3 of her ICU stay and blood cultures and wound swab that were sent in the Emergency Room have grown methicillin resistant staph aureus (MRSA). The surgical consultant recommends surgical debridement and urgent colostomy for source control and to prevent further wound contamination
7. I would initiate Renal Replacement Therapy if Ms. Y becomes anuric and acidotic	Strongly agree	24	13	0.004	0.081	13	0	0	0.003
	Somewhat agree	11	11			13	12
	Neither	2	12			8	10
	Somewhat disagree	2	3			5	14
	Strongly disagree	0	0			1	3
8. I would encourage Ms. Y to undergo the recommended surgical interventions immediately	Strongly agree	12	5	0.02	0.341	7	3	0.851	1
	Somewhat agree	19	16			8	14
	Neither	3	12			18	17
	Somewhat disagree	4	6			7	3
	Strongly disagree	1	0			0	2
9. I would intubate Ms. Y if she has a respiratory arrest	Strongly agree	30	5	0	0	0	0	0.011	0.203
	Somewhat agree	7	9			0	1
	Neither	1	13			0	3
	Somewhat disagree	0	5			5	10
	Strongly disagree	1	7			35	25
10. I would perform CPR if Ms. Y became pulseless	Strongly agree	29	5	0	0	0	0	0.001	0.014
	Somewhat agree	7	9			0	1
	Neither	2	15			0	3
	Somewhat disagree	0	4			2	10
	Strongly disagree	1	6			38	25

The effect of patient elected code status on medical decisions outside of cardiopulmonary resuscitation (Tables 2 and 3 and Supplemental Tables 2 and 3)

Our survey demonstrated multiple instances when patient code status alone appeared to affect the decisions of both groups of trainees in the provision of care unrelated to CPR and intubation.

In Vignette #1, American and French trainees were less likely to agree with offering dialysis for anuria and acidosis (corrected p-value 0.012 and 0) when Ms. L decompensates with DNR/DNI status. French trainees were less likely to agree with central line placement when peripheral IV was unavailable (adjusted p-value 0.026) with DNR/DNI status, while code status did not affect American trainees' decisions. However, DNR/DNI status made American trainees less likely to agree with surgical consultation for colectomy (adjusted p-value 0.049) or GI consultation for hematemesis (adjusted p-value 0.008). American trainees were more likely than French trainees to start peripheral pressors for refractory hypotension, regardless of code status (adjusted p-value 0). With Full Code status, American trainees more strongly favored surgical consultation for colectomy than French trainees (p-value 0.0007).

In Vignette #2, Ms. Y's code status did not significantly impact therapies unrelated to CPR and intubation among American trainees. French trainees were less likely to offer peripheral pressors for hypotension (adjusted p-value 0.021), ICU transfer for shock after fluid resuscitation (adjusted p-value 0.027) and GI consult for hematemesis (corrected p-value 0.002) when Ms. Y had DNR/DNI status. In the ICU, French trainees were less likely to offer transesophageal echocardiogram for MRSA bacteremia or initiate renal replacement therapy for anuria/acidosis (adjusted p-value 0.001 for both).

American trainees were more likely to strongly agree with fluid resuscitation with DNR/DNI status (adjusted p-value 0.003), and peripheral vasopressors initiation for hypotension regardless of the code status (adjusted p-value 0.046 Full Code; 0 DNR/DNI). Americans more strongly favored ICU transfer for hypotension, central line placement, and GI consultation with DNR/DNI status (adjusted p-value 0).

In the ICU, French trainees were less likely to administer second vasopressors for instability (adjusted p-value 0), stress dose steroids with DNR/DNI, renal replacement for anuria/acidosis with DNR/DNI (adjusted p-value 0.003), and GI consultation regardless of code status (adjusted p-value 0.025 Full Code, 0.002 DNR/DNI).

Discussion

In this transatlantic survey, we demonstrated that patient code status alone affects the medical decision-making of both French and American critical care trainees with regard to provision of therapies unrelated to intubation and cardiopulmonary resuscitation. As demonstrated previously in studies of American medical, surgical, and emergency medicine physician trainees [12–14], both American and French critical care trainees were generally respectful of the patient’s desired code status with regard to providing or withholding intubation and cardiopulmonary resuscitation, but seemed to interpret a DNR/DNI code status to mean that the patient desirous of less aggressive care overall.

Our original hypotheses were that American trainees were likely to interpret DNR/DNI status as signifying a desire for less aggressive care, and that French trainees were likely to act directly against patient wishes with regard to provision of CPR, and these theories were only partially supported by our evidence. In fact, in the case of Ms. L, the 70 year old woman with Clostridium difficile infection, both groups seemed to extrapolate that DNR/DNI status reflected a desire for less aggressive care in general, whereas for Ms. Y, the 38 year old woman with MS who is bedridden and presented with a wound infection, only the French group seemed to maintain this assumption. One possible explanation is that Ms. Y’s progressive chronic disease, with which she is likely very familiar, may have created a cognitive bias among physicians in the survey, leading to greater acceptance of her autonomy in decision-making. With regard to our hypothesis that the French trainees would be more paternalistic in their decisions around providing CPR and intubation, we found instead that they were statistically more likely than not to be respectful of patient code status although they were less polarized in this than their American counterparts.

Our study suggests that despite potentially conflicting ethical codes, the concrete medical practice of French and American trainees is not actually that different. Both groups appear to assume that a DNR/DNI status indicates that patients may desire less aggressive care overall, and both groups are statistically more likely than not to respect a patient’s elected code status. The difference was mostly in how decisive the decisions were, indicating that the French are not as paternalistic as they might desire to be, while the Americans may be more paternalistic than they admit to being. Notably, in each vignette, respondents are given the opportunity to discuss goals of care with each patient and their healthcare proxy while the patient is on the hospital ward and again after transfer to the ICU. There was no statistical difference between French and American trainee agreement with the decision to discuss goals of care with the patient and their healthcare proxy, regardless of patient code status or location, reflecting that in the midst of a rapidly-evolving hospital course, both groups desire to take into account the input both the patient and their designated decision-maker.

Apart from these empirical findings, the interpretation of our results should be situated within broader ethical and cultural contexts. Our explanation for our findings is twofold: From the European perspective, over the last two decades, the French seem to be moving away from paternalism. Recent developments in France illustrate that the traditional emphasis on beneficence and physician stewardship is beginning to evolve toward a greater recognition of patient autonomy. The Leonetti (2005) and Claeys-Leonetti laws [23, 24] (2016) laid early foundations by prohibiting unreasonable obstinacy in care, strengthening advance directives, and authorizing continuous deep sedation until death. More recently, in May 2025, the French National Assembly (First Chamber) approved a landmark bill legalizing medical aid in dying under strict conditions for competent adults with incurable illnesses who experience intolerable suffering [25]. This bill is now being examined by the Senat (Second chamber). While narrower in scope than the more permissive frameworks of Belgium [26] or the Netherlands [27] this legislation represents a profound cultural shift. For the first time, French law formally enshrines a patient’s autonomous choice to hasten death, reflecting ethical and societal pressure to align end-of-life care with individual patient values rather than physician-determined assessments of beneficence. These reforms suggest that the traditional paternalistic model in French critical care ethics is being recalibrated, with autonomy increasingly recognized as a legitimate and necessary counterweight to beneficence.

From the American perspective, we believe our findings reflect an unspoken acknowledgement that the American ideal of having patient autonomy be the sole or even major directing factor in the care provided may simply not be achievable. For instance, even when patients are appropriately deemed decisional, meaning that they have the ability to understand factual information and communicate their decisions, the capacity for decision-making is not sufficient for autonomy. Autonomy assumes an informed patient who fully understands the implications of available treatment choices and can identify those most aligned with their values and goals [28], and even in the absence of the delirium that is often pervasive in the intensive care unit [29], it is not possible to estimate the influence that critical illness can have on a given patient’s (or their family’s) emotional state. Studies have shown that fear increases the perceived consequence of negative outcome, sadness increases risk tolerance, anxiety increases risk aversion, and negative mood impairs logical reasoning processes [30, 31]. The American physician in this situation can find themselves having to choose between providing what they believe to be inappropriate, futile, or even harmful care [32] at the directive of a patient or family, or imposing their own values on the patient or family in a practice of paternalism that they have been culturally mandated to work against.

Beyond the survey results themselves, these findings provide an opportunity to reflect on whether the future of critical care ethics may lie not in choosing between autonomy and paternalism, but in integrating both traditions. In order to avoid abandoning patients and their families by shifting the full weight of life-and-death decisions onto them when they may be exhausted, frightened, or incapable of fully informed deliberation, respect for autonomy must remain a central commitment. Yet beneficence—properly understood as an expression of professional responsibility rather than physician authority—must also be preserved to protect patients and families from the unbearable burden of impossible choices. In this sense, the evolving French legal framework on assisted dying highlights a broader convergence: autonomy and beneficence need not be antagonistic principles, but can be woven together into an ethic of care that both honors patient values and safeguards patients from harm. Increasingly, scholars of medical ethics have argued for a model of relational autonomy—an approach that recognizes patients as embedded in networks of family, culture, and professional guidance, rather than as isolated decision-makers [33]. Physician recommendations integrate an understanding of medical feasibility with patient values, including their preferences regarding hospitalization, functional outcomes, and uncertainty. Accordingly, respect for autonomy is operationalized through shared decision-making within clinically realistic options, rather than the presentation of an unrestricted set of choices. As such, there is a need for a more personalized approach to medical ethics, mirroring the evolution of personalized medicine in technical domains of critical illness such as ventilation strategies [34, 35], pain management [36], nutrition [37], and even hemodynamic monitoring [38]. Just as no single therapeutic protocol suits every patient, it is unlikely that a one-size-fits-all ethical framework can adequately address the diverse needs, values and expectations of critically ill patients and their families. A rigid or uniform approach risks either overwhelming vulnerable individuals with decisions they are not prepared to make or marginalizing their voice in the care they receive. Instead, critical care physicians must navigate these complexities with flexibility and humility, adapting their ethical approach to the unique preferences and capacities of each patient and family. By fostering true shared decision-making, tailored to the specific context of the individual, we can better uphold the principles of beneficence, justice, and respect for autonomy to provide the best path forward, away from harm.

Strengths and limitations

Our study has several strengths. It is the first study to evaluate specific medical decisions between French and American critical care training physicians. We supply national data from both countries that is diverse with respect to geographic location, and level of training. Our study also has limitations. Patients were hypothetical and our respondents were unable to clarify patient wishes beyond the instructions provided by the survey. Therefore, our study reflects a theoretical effect on physician practice. However, responses to clinical vignettes have previously been shown to correlate with actual practice [39, 40]. Additionally, although we had enough respondents to be able to reach statistical significance, our response rate was low in France (19%) and cannot be calculated in the United States, as we do not know how many program directors passed along our survey to their trainees.

Moreover, using the first letter of the last name might have introduced a structural imbalance in the dataset because different cultures have a different distribution of the first letter of the last name.

Finally, another potential limitation relates to differences in age and clinical experience between French and US trainees. Although these differences may have contributed to some of the observed findings, particularly given the less standardized approach to code-status procedures in France, current evidence does not support physician age or seniority as independent determinants of end-of-life decision making, and the influence of these factors therefore remains speculative [41–43].

Conclusion

The present transatlantic study highlights the influence of patient-elected code status on critical care trainees' medical decisions. It suggests a shift toward integrating patient autonomy with professional responsibility to prevent harm, emphasizing a personalized, relational ethical approach. Overall, the findings call for flexible, context-sensitive ethical frameworks that support shared decision-making tailored to individual patient needs.

Abbreviations

US

United States

ICU

Intensive care unit

DNR

Do not resuscitate

DNI

Do no intubate

IV

Intravenous

CPR

Cardiopulmonary resuscitation

MRSA

Methicillin-resistant staphylococcus aureus

Authors’ contributions

**E. Stevenson**: Creation of the vignettes, distribution of the survey in the United States, writing the manuscript, and compiling the bibliography. **S. Boccand**: Development of the survey in LimeSurvey, assistance with survey distribution in France, data collection, creation of tables and figures, writing the manuscript, and compiling the bibliography. **A. Lautrette**: Manuscript review, providing corrections, and compiling the bibliography. **M. Giry**: Distribution of the survey and provision of access to the data collection site (LimeSurvey). **M. Fosset**: Statistical analysis, interpretation of the data. **B. Jung**: Management and organization of work progress, creation of the vignettes, assistance with survey distribution in France, input on manuscript writing, and compilation of the bibliography. All authors read and approved the final manuscript.

Funding

There are no sources of funding to declare for this research.

Data availability

The data were collected using LimeSurvey (Supplemental Material), with access provided by the French Young Intensive Care Physicians Association in France and American Program Directors were contacted via email addresses listed on the Society of Critical Care Anesthesiologists (SOCCA) and FREIDA AMA Residency & Fellowship Programs databases.

Declarations

Ethics approval and consent to participate

This project was validated by the institutional review board of Beth Israel Deaconess Medical Center and the university hospital center of Montpellier and consent to participate was waived. This project was performed in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.