Patients’ psychosocial attributes and aggressiveness of cancer treatments near the end of life

Abstract

Background

While the use of chemotherapy near the end of life (EOL) has been identified as a relevant criterion for assessing quality of cancer care and has been estimated as non-beneficial, a trend of aggressiveness in cancer care during the last period of life remains. Both patients’ sociodemographic characteristics and physicians’ practice setting are associated with this use. The role of patients’ psychosocial characteristics has however been understudied. The objectives were to study oncologists’ intention to recommend chemotherapy or therapeutic abstention in an EOL patient’s case and to examine the factors associated with this decision.

Methods

A clinical vignette-based questionnaire survey was conducted. While the case presented to the participating oncologists of a patient with EGFR-mutated lung cancer, progressing after osimertinib, ECOG 3, with leptomeningeal disease (N = 146), was strictly equivalent in terms of medical aspects and age, 4 patients’ non-medical characteristics were manipulated: gender, marital status, parenthood, and psychosocial characteristics (“nice” patients, patients “making strong demands,” or control patients).

Results

77.4% of the oncologists surveyed stated that they would recommend chemotherapy in this situation. Only scenarios with nice patients or patients making strong demands were associated with less recommendation of chemotherapy (70.8% for the nice/making strong demands scenarios together vs 87.7%, for the control scenario P = .017). Medical oncologists with previous experience of similar cases were also less keen to recommend chemotherapy (73% vs 100%, P = .007). Of the 76.7% of respondents declaring that they would think of other therapeutic options, 49.1% mentioned “other treatments” without mentioning palliative care.

Conclusion

Developing physicians’ awareness of the psychosocial aspects at stake in their medical decisions in these sensitive situations may improve EOL care.

Implications for practice.

The use of anticancer treatments near the end of life (EOL) has largely been estimated as non-beneficial and shown to be associated with patients’ sociodemographic characteristics. This study assessed, through a clinical vignette methodology, the associations between patients’ psychosocial attributes and chemotherapy recommendations in an EOL patient’s case. Situations leading to more socio-affective closeness with the patient (where the patient’s psychosocial attributes were salient) were associated with less recommendation of chemotherapy. These results have directly applied perspectives in the development of training targeting oncologists’ awareness regarding the psychosocial aspects at play in their medical decision-making, especially in emotionally adverse situations such as EOL care.

Introduction

The use of anticancer treatments near end of life (EOL) has been identified as a relevant criterion for assessing the quality of care of patients with cancer^1,2 and has largely been estimated as non-beneficial considering classical endpoints in medical oncology.^3-5 Indeed, the use of anticancer treatments in patients with poor performance status has a limited impact on their overall survival, increases care costs, and harms patient quality of life because of the numerous side effects of chemotherapy.^6,7 However, several recent studies have revealed an increasingly aggressive trend in cancer care during the last months and weeks of life, such as the high use of cancer therapies.^8-10 For example, chemotherapy administration 1 month before the patients’ death varies between 5% and 55% depending on studies.^11-13 These data contrast with calls, which have been emphasized in the different national cancer plans in France, to support the integration of palliative care in the cancer care routine.¹⁴ Indeed, a substantial body of literature has shown the benefits of Early Palliative Care for quality of life and clinical outcomes such as symptom intensity^15,16 while it is still underutilized and the time from referral to death remains short.¹⁷

As in other critical decisional contexts where uncertainty is strong and medical knowledge is lacking, non-medical factors are associated with these prescriptions. Non-medical influences on medical decision-making including patients’ non-clinical factors, have been widely documented in the Human and Social Sciences literature with a major focus on patient sociodemographic attributes (eg, ethnicity, socio-economic status, etc.).^18-21 This topic has mainly been addressed in situations where medical, social, and moral judgments can merge (eg, stigmatized pathologies such as HIV or mental illnesses). Indeed, these contexts are preferred fields for studying the psychosocial aspects involved in medical practices and in the patient-physician relationship.

For similar reasons, research has focused on advanced cancer contexts when therapeutic options diminish. Among the factors associated with chemotherapy prescription near EOL, two French retrospective and registered-based studies showed that patient age and gender as well as the type of institution might influence chemotherapy rates during the 3 months before death: young patients, males and patients receiving treatment in a comprehensive cancer center or private hospital were more likely to receive chemotherapy during the 3 months before their death.^22,23

Specifically, regarding patients’ non-medical characteristics, qualitative studies have shown that multiple non-clinical factors (eg, life circumstances, family related obligations, etc.) were mentioned by physicians as criteria that intervened in their decisions.^24-27 Important in that they reveal oncologists’ subjective perceptions of factors influencing their prescriptions in EOL situations, these studies nevertheless do not provide information about variables actually affecting oncologists’ intention to prescribe chemotherapy near EOL. Through different methodological designs, studies conducted in the field of oncology have shown the influences of sociodemographic (eg, ethnicity, marital, and socio-economic status, age, etc.) and “psychological” attributes (eg, patients’ presentation style, mental health, etc.) on decision-making.^28-34 These psychosocial characteristics, less studied, are attracting increased attention and are worth exploring to increase the variables that can be found in patients’ records and thus deepen our understanding of oncologists’ decision-making in these critical clinical situations. To do so, the comprehensive and phenomenological approach of social representations,³⁵ developed in social psychology, which focuses on how people make sense of the world based on shared knowledge socially produced and used to direct their actions and communications appears as a relevant theoretical perspective. Here, patients’ psychosocial characteristics are considered as a form of knowledge shared and drawn on by oncologists as efficient categories for guiding their practices when facing difficult decision-making.³⁶ Studying oncologists’ decision-making regarding chemotherapy near EOL with this approach involves considering what oncologists would do rather than what oncologists should do and attempting to understand why they would do it this way. More broadly, the psychosocial perspective³⁷ adopted in this work leads us to consider the social context in which medical-decision making is anchored to analyze the psychosocial processes underlying it. The social context is here understood as both the macro-context (including the cultural and ideological climate surrounding oncology such as the current valorization of autonomy in patient care) and the micro-context (including the interpersonal and intergroup relationships between the physician and the patient).

Based on previous studies showing the emergence and influence of patients’ non-medical characteristics on critical decision making in oncology, the objectives of the present study were: (1) to study oncologists’ intention to recommend chemotherapy or therapeutic abstention in an EOL patient’s case; (2) to examine the factors associated with this medical decision.

Materials and methods

From the perspective of a methodological triangulation,³⁸ we made the choice to focus our attention on patients’ psychosocial variables, understudied and identified in a previous fieldwork combining multidisciplinary team (MDT) meeting observations and oncologists’ semi-directed interviews to refine the understanding of the psychosocial processes underlying their use by oncologists.

Study design

A clinical vignette-based questionnaire survey was conducted.^39-41 Frequently used in the medical field, the clinical vignette method consists of presenting physicians or medical students with clinical cases written to appear as close as possible to real-life situations of care and to ask them how they would manage them. In this study, this tool was used for its advantage of isolating and testing the influence of a few specific variables on medical recommendations regarding therapeutic strategy. Specifically, we chose a “between-subjects research design” where participants are randomly assigned to different vignette groups, such that all participants within a group are presented with the same single vignette and participants in different groups are presented with different vignettes.⁴¹ The medical situation depicted in the vignette was co-designed by a pluri-disciplinary team composed of medical oncologists, palliative care physicians, and social science researchers (social psychologists, statisticians). We designed the medical situation to maximize the uncertainty of initiating or not initiating an anticancer treatment: no guidelines and few scientific data, questionable or with a low level of evidence, were available to base a decision on.

Thus, the scenario described a patient with an advanced lung adenocarcinoma, EGFR mutated, with leptomeningeal disease progression after a first line of osimertinib (detailed in Supplementary Appendix 1). While the cases were strictly equivalent in terms of medical aspects and age (a 42-year-old patient), 4 patients’ non-medical characteristics were manipulated: gender (x 2), marital status (x 2), parenthood (x 2), psychosocial characteristics (x 3), corresponding to 24 different possible scenarios (Figure 1). The psychosocial characteristics manipulated in the scenario (nice patient or patient with strong demands) were chosen based on real patients’ characteristics mentioned during in vivo situations of medical decision-making observed in a previous qualitative study.⁴² Depicted by oncologists as falling under the category of patients’ “psychological portrait,”³⁸ these categories actually cover a plurality of aspects combining patients’ social, clinical, and psychological characteristics which are eminently social since they have a shared and socially produced meaning. The meaning-making of these categories is also anchored in the socio-affective and relational climate developed through patients’ and physicians’ relationships. These organized characteristics on the basis of which oncologists infer patients’ personality are associated with specific expectations in terms of patients’ behaviors or wishes and appear for this reason as efficient categories, used by oncologists in professional situations to distinguish between and classify patients and guide their practices. The choice of these different characteristics was underpinned by the social values differently attributed to them. As an example, the profile of a young married mother is the prototypical figure of a socially valuable patient.³⁶ In that sense a more “active” treatment strategy was expected to be recommended for married patients and patients with children.

Figure 1.

Study design.

Outcomes measured

The principal outcome measured was the oncologists’ recommendation of chemotherapy or withholding of a specific treatment without prior knowledge of the patient’s preferences and apart from the initiation of palliative care. Participants were asked to check one of these two options.

The other outcomes measured were:

• - The oncologists’ recommendation of another treatment option that could be offered to the patient in their opinion through a “yes/no” question. The respondents checking “yes” were then requested to describe the strategies considered in an open-ended question. The analysis of these strategies was based on a researchers’ triangulation procedure involving clinicians and social scientists. This resulted in the quantitative coding of this outcome into 3 modalities: “exclusive palliative care,” “palliative care combined with other treatments” or “other exclusive treatments.”

• - The credibility of the clinical vignette presented, checked through a “yes/no” question regarding oncologists’ experiences of similar clinical cases.

• - The correct induction of patients’ psychosocial characteristics, assessed through 4 questions based on an Osgood scale ranging from (1) not at all to (5) high, where respondents were asked to rate their perception of the patient’s sympathy, demand, boredom and attachment.

• - Oncologists’ sociodemographic (gender, age, marital, and familial situation) and professional (number of years of practice, main place of work) characteristics. Senior oncologists were identified as those having at least 10 years of practice after completing a medical degree (MD). Hospitals were grouped according to their type: comprehensive cancer centers/private hospitals, university hospitals, other settings.

Data collection and analysis procedures

A link to an online survey conducted with the LimeSurvey tool was shared with French oncologists through different professional societies specializing in oncology between October 2021 and July 2022. The circulation of the online survey aimed to target different types of oncologists’ profile according to their years of experience (junior/senior), gender, hospital types (cancer centers and university hospitals), and regions of practice. Based on the need for only one link to be sent to all practitioners, and for randomization to be carried out when the questionnaire was administered, it was therefore programmed by a script inside the LimeSurvey program, as a hidden question. This hidden question assigned each new participant an incremental scenario so that each scenario was seen by the same number of practitioners and was completely randomly distributed among the respondents. Subsequent differences in numbers of participants by scenarios emerged after the data management process, that is, removing incomplete responses and/or practitioners other than oncologists. Insofar as our analyses concerned variables that included several scenarios in each case, our conditions for applying the statistics were sufficient to ensure their validity. Data were anonymous and the Ethical Committee of Aix-Marseille University approved the protocol (IRB00014113, approval number 2021-04-08-07).

We first performed ANOVAs to check oncologists’ perception of the patient’s sympathy, demand, and attachment according to the scenarios they were exposed to. Then, the frequencies of chemotherapy recommendation by oncologists (principal outcome) were compared depending on the scenario they were exposed to (4 variables described above) and according to their socio-demographic and professional characteristics. Chi-squared or Fisher’s exact tests were used for categorical variables and ANOVAs for continuous variables. A P-value < .05 was considered as statistically significant. Analyses were performed using IBM SPSS version 27.SPSS.

Results

Characteristics of the study sample and credibility of the vignette presented

Our sample was made up of 146 oncologists practicing in the 13 different French regions who responded to the questionnaire. 56.2% were women, and their mean age was 43.5 years (min: 27; max 67; SD = 9.9). More than half of the oncologists (53.4%) were working in comprehensive cancer centers/private hospitals compared to 41.1% in university hospitals and 5.5% in other settings and 54.1% were senior oncologists.

In total 82.9% of the oncologists surveyed had already encountered a similar case in their careers, which supports the relevance of the clinical vignette developed for the survey. The oncologists’ perception of the patient depicted in the vignette varied according to the scenario they were exposed to. The “nice” patient was seen as “nicer” (4.37; SD = 0.75) and more endearing (4.35; SD = 0.74) than the patient making strong demands (3.97; SD = 0.97 and 4.03; SD = 0.79) and the control patient (3.61; SD = 0.80 and 3.77; SD = 0.85, P < .001 and P = .001, respectively). The patient making strong demands was seen as making more demands (4.45; SD = 0.60,) than the “nice” patient (3.33; SD = 0.91) and the control patient (3.53; SD = 0.80, P < .001), confirming that the psychosocial characteristics manipulated in the scenario were actually inducing the expected characteristics.

Recommendation of chemotherapy and associated non-medical factors

A majority of the oncologists (77.4%) stated that they would recommend chemotherapy in this situation of patient with EGFR-mutated lung cancer, progressing after osimertinib, ECOG 3, with leptomeningeal disease (Table 1). Physician’s characteristics such as gender, age, and type of center were not associated with chemotherapy prescription, with very similar rates of chemotherapy (around 75%-80%).

Table 1.

Frequency of a recommendation of chemotherapy for the critical clinical case presented and associated factors.

	Recommendation of chemotherapy
	Total n = 146	No 33 (22.6)	Yes 113 (77.4)	P-value
Patient depicted in the scenario
Gender
Man	88 (60.3)	21 (23.9)	67 (76.1)	.654
Woman	58 (39.7)	12 (20.7)	46 (79.3)
Marital status
Single	63 (43.2)	15 (23.8)	48 (76.2)	.761
Married	83 (56.8)	18 (21.7)	65 (78.3)
Familial status
Childless	60 (41.1)	12 (20.0)	48 (80.0)	.530
With children	86 (58.9)	21 (24.4)	65 (75.6)
Psycho-social characteristic
Nice	51 (35.0)	16 (31.4)	35 (68.6)	.049
Making strong demands	38 (26.0)	10 (26.3)	28 (73.7)
Control	57 (39.0)	7 (12.3)	50 (87.7)
Scenario
Nice/making strong demands	89 (61.0)	26 (29.2)	63 (70.8)	.017
Control	57 (39.0)	7 (12.3)	50 (87.7)
Oncologist characteristics
Gender
Man	64 (43.8)	19 (29.7)	45 (70.3)	.071
Woman	82 (56.2)	14 (17.1)	68 (82.9)
Age mean (SD):	43.6 (9.9)	44.2 (10.7)	43.3 (9.7)	.652
Min-max	27-67	29-66	27-67
Senior oncologist
No	67 (45.9)	14 (20.9)	53 (79.1)	.650
Yes	79 (54.1)	19 (24.1)	60 (75.9)
Work in a Comprehensive Cancer Center/private hospital
No	68 (46.6)	13 (19.1)	55 (80.9)	.347
Yes	78 (53.4)	20 (25.6)	58 (74.4)
Experience of similar cases
Yes	121 (82.9)	32 (26.4)	89 (73.6)	.007*
No	19 (13.0)	-	19 (100.0)
Missing	6 (4.1)	1 (16.7)	5 (83.3)

^*Fisher’s exact test (calculated excluding missing values).

At patient-level, sociodemographic characteristics in the scenario such as gender, marital status, and parenthood were not associated with chemotherapy prescription (Table 1).

Only the psychosocial characteristics of the patients manipulated in the scenarios were associated with the intention to recommend chemotherapy. The oncologists exposed to a “nice” patient or a patient “making strong demands” version of the scenario declared that they would recommend chemotherapy less frequently than the oncologists exposed to the control (with no information about the patients’ psychosocial characteristics) version of the scenario (70.8% for the nice/“making strong demands” scenarios together vs 87.7% for the control scenario, P = .017). This impact on decreasing chemotherapy recommendation appeared significant either when studying “nice” or “making strong demand” variables separately, or when grouping these 2 items together.

Second, medical oncologists with previous experience of similar cases recommended chemotherapy prescriptions less frequently in this clinical situation (73% vs 100%, respectively, P = .007).

Other treatment options considered

First, 76.7% of the respondents declared that they would think of other treatment options in this case with 49.1% mentioning “other treatments” alone, without mentioning palliative care (Table 2). These treatments included radiation therapy (stereotaxic or whole brain), or systemic medical treatments composed of targeted therapy, antiangiogenic, or immunotherapy. It is noteworthy that some oncologists suggested invasive and aggressive procedures such as intra-thecal injection of chemotherapy or corticoids, either alone (12.5%) or combined with other treatments (9.9%).

Table 2.

Other treatment strategies mentioned by the respondents coded in three categories

Three other therapeutical strategies coded	Details	N	%
Exclusive palliative care	Supportive and palliative care	19	17.0
	Supportive and palliative care + Symptomatic medical care	6	5.4
	Symptomatic medical care	2	1.8
	Total	27	24.1
Palliative care combined with other treatments	Supportive and palliative care + Oncological medical care	12	10.7
	Supportive and palliative care + Intrathecal treatment	8	7.1
	Supportive and palliative care + Oncological medical care + Intrathecal treatment	3	2.7
	Supportive and palliative care + Symptomatic medical care + Intrathecal treatment	2	1.8
	Supportive and palliative care + Symptomatic radiotherapy	2	1.8
	Supportive and palliative care + Symptomatic medical care + Symptomatic radiotherapy	2	1.8
	Supportive and palliative care + Symptomatic radiotherapy + Intrathecal treatment	1
	Total	30	26.8
Other exclusive treatments	Oncological medical care	15	13.4
	Intrathecal treatment	14	12.5
	Symptomatic radiotherapy	7	6.3
	Oncological medical care + Intrathecal treatment	7	6.3
	Oncological medical care + Symptomatic radiotherapy	5	4.5
	Intrathecal treatment + Symptomatic radiotherapy	5	4.5
	Oncological medical care + Symptomatic medical treatments	1	0.9
	Oncological medical care + Symptomatic radiotherapy + Intrathecal treatment	1	0.9
	Total	55	49.1

Concerning palliative care, only 50.9% of oncologists mentioned it spontaneously, either combined with some anticancer treatments (26.8%) or combined with exclusive symptomatic medical treatments, such as corticoids or analgesics (24.1%).

Discussion

This clinical vignette-based research aimed to study French oncologists’ intention to recommend chemotherapy in a critical patient’s case near EOL and to examine the non-medical factors associated with such recommendation. It was developed through a socio-representational approach involving not attempting to evaluate oncologists’ practices based on predefined standards judged relevant to do so but shedding light on the psychosocial processes underlying their treatment suggestions. In this study, chemotherapy remains the preferred option considered by oncologists in the given situation of a patient near the end of life. The fact of recommending the withholding of a specific treatment was positively associated with oncologists presented with clinical cases including patients’ psychosocial characteristics such as being nice or making strong demands. Therapeutic abstention was also associated with a previous experience of a similar case.

The level of aggressiveness of oncological treatments proposed by oncologists in the survey was high, with a low rate of palliative care referral (around 50%), a high rate of cytotoxic chemotherapy (around 75%), and a high rate of intrathecal chemotherapy (around 25%). However, in the case of a poor performance status patient (PS3) with leptomeningeal disease (LMD) from non-small cell lung cancer (NSCLC), most cancer society’s recommend exclusive palliative care. As an example, the US National Comprehensive Cancer Network (NCCN 2023. version 1.2023) guidelines recommend best supportive care for patients with limited performance status and major neurological symptoms. The European Society for Medical Oncology (ESMO) guidelines state that poor performance status (PS3-4) patients should be offered the best supportive care.⁴³ Indeed, a study has described how quality of life in patients with end-stage cancer is not improved and can be harmed, by chemotherapy use near death.⁴⁴ Concerning intrathecal chemotherapy, no randomized trials have ever proved its benefit in NSCLC (despite the use of numerous molecules), but associated adverse events are high, notably pain and immunosuppression.⁴⁵ Similarly, a retrospective study of LMD secondary to NSCLC showed no overall survival benefit with whole-brain radiotherapy.⁴⁶

In our results, many treatment combinations were suggested by oncologists, combined or not with palliative care introduction. The topic of whether to prolong or end the treatment now translated as “one more chemo or one too many” is not new in oncology⁴⁷ and continues to fill medical editorials.^13,48 However, it has been reshaped by the continuous development of new therapeutic options based on the expansion of scientific knowledge. This offers oncologists the possibility of negotiating creative proposals even late in the progression of the disease in the so called bricolage phase⁴⁷ more widely embedded in a socio-cultural and ideological context where hope, optimism and fighting spirit remain key values in cancer care.⁴⁹

Contrary to studies analyzing variation in cancer care according to patients’ non-medical characteristics,^29,31,32,50 patient gender, marital, or parental status were not found to be associated with oncologists’ recommendation of treatments in the present study. However, contrary to our expectations, their psychosocial characteristics such as being nice and making strong demands were associated with the recommendation of less aggressive treatments. In the control version of the scenario, patient age was the only indicator that could have a social meaning on top of its clinical significance. Indeed, young patients have been shown to arouse a more emotional reaction due to the social value associated with young age³⁶ and to receive more aggressive treatments.²² The patient’s psychosocial characteristics present in the two other scenarios appear to “neutralize” or transcend the effect of age. Even if the result could seem more surprising in the scenario presenting the patient making strong demands regarding the influence of patients’ preferences on medical decisions,^31,32 the methodological operationalization of the nice patient and one “making strong demands” have in common the fact that they highlight the socio-affective side of the medical relationship. More precisely, in the “nice” version of the scenario, this socio-affective side reflects the relational climate developed between the oncologist and the patient based on common interests (regarding food here) and amplifying the possibility of social identification between the patient and the physician. This aspect is explicitly translated by the terminology “feelings,” frequently used by oncologists themselves in MDT meetings and in the semi-directive interviews.³⁶ The “patient making strong demands” version of the scenario operationalized the issue of hope through the patient’s expression of the wish to still be alive for a loved one’s birthday taking place a few months later. Here, hope is rooted in the social affiliation of the patient (familial or friendly) which makes him/her appear as concretely inscribed in social relations with others, such as participating in social life and giving him/her a significant social value. In both cases, through the patient’s relationship with the physicians or through his/her relationships with significant others, these socio-affective dimensions make the patient salient as a “subject” rather than an “object” reduced to his/her disease and may explain oncologists’ inclination for a therapeutic approach based on patient quality of life. Indeed, this could lead to more social identification with and projection toward the patient and may explain the recommendation of what can be viewed as a more “reasonable” strategy according to medical standards. In these situations of socio-affective closeness, a perspective shift “from cure to care” seems to be in play. Moreover, in view of the fact that “worries about withdrawal or inappropriate continuation of cancer treatment”⁵¹ (p. 1736) have been mentioned as stressful situations in a literature review focused on oncologists’ mental health and that links have been shown between the experience of the emotionally powerful death of a patient and the physician following clinical practice,⁵² having had a previous similar experience and being confronted with a nice patient or one making strong demands may have led oncologists to decisions they were less likely to regret. Deeper analyses of oncologists’ perception of the patient in the different scenarios as well as their explanations regarding the motivation underlying their recommendations are needed to examine these interpretations.

Limitations

Although this study questions, through an original methodological design, the non-medical determinants of chemotherapy prescription near the end of life, it presents some limitations. Inherent to the use of clinical vignettes, the scenarios presented to the oncologists are simplified versions of a patient’s clinical case and do not allow the complexity of the multiple elements at play in these situational contexts to be addressed. This complexity could, for example, have been approached by manipulating other patient characteristics such as their ethnicity or by extending the scenarios to the wishes of other important actors in these decisions, namely the patients’ caregivers. However, the usual low rate of responses from physicians in surveys compared to the general population⁵³ and the subsequent small sample of participants do not allow the variables tested to be multiplied. Their limited time availability also constrains the duration of the questionnaire and thus limits the possibility of many outcomes to be measured. For these reasons and regarding the importance of physician’s recommendations, despite the push towards shared-decision making,⁵⁴ we decided to focus in this vignette study on the oncologist’s perspective while patients’ values would have been worth taking into account. In the same way, some important themes directly connected to the decision-making question such as the one of the communication of bad news or providing goal concordant care could not be included in the study. This gap between the study design and the reality of medical decision-making near EOL seems nevertheless nuanced by the ability that the respondents showed to place themselves in the depicted situations as testified by the large majority of oncologists declaring having encountered a similar situation in their daily clinical practice.

A second limitation is that statistical analyses focusing on the influences of patients’ non-medical characteristics on treatment recommendation were conducted taking into account each of the 4 variables manipulated separately. The possible interaction between these variables within a scenario could not be analyzed due to the size of the study sample. The qualitative analyses of the open questions included in the questionnaire about oncologists’ perception of the patient depicted in the scenario will help us to understand which patient characteristics attracted oncologists’ attention and the overall impression that the patient’s presentation gave them according to the different scenarios.

Applied perspectives and further research directions

Research focusing on healthcare professionals’ perceptions of non-clinical criteria intervening in their medical decision-making has shown that physicians seem aware of the intervention of such factors in this area.^24-27 Moreover, these aspects of clinical encounters and their particular nature in oncology where oncologists are facing emotionally adverse situations such as patients’ unrealistic demands, uncertainty about introducing the right treatment and patient death have been shown to be linked to the poor mental health of oncologists.⁵¹ Engaging oncologists in reflexivity regarding the intervention of non-medical aspects in their practices appears to be an interesting applied perspective, as this study is part of a wider ongoing action-research project being implemented throughout our institutions. Clinical vignettes have been used in pilot training programs as an intervention tool aimed at raising oncologists’ awareness of these psychosocial aspects of their professional practices.

Diversifying the variables being tested and notably introducing the wishes of the multiple actors involved in these therapeutic strategy decisions into studies designed with clinical vignettes would constitute a fruitful direction for research. Exploring how these therapeutic strategies are negotiated in dedicated consultations between the different individuals involved and correlating their perspectives is another interesting direction currently developed in an ongoing mixed-design study combining observations of consultations and patient-oncologist questionnaires.

In conclusion, this work has demonstrated that patients’ psychosocial variables are associated with chemotherapy prescription and shed light on the importance of considering the social branding of patients from a broader angle while analyzing physicians’ medical decision-making. Altogether, we think that developing physicians’ awareness regarding the psychosocial aspects at play in their medical decision-making in these sensitive situations might improve EOL oncological care.

Author contributions

Léa Restivo (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Writing—original draft) Philippe Rochigneux (Formal analysis, Methodology, Writing—original draft) Anne-Déborah Bouhnik (Data curation, Formal analysis, Methodology, Writing—original draft) Thomas Arciszewski (Software) Aurélie Bourmaud (Methodology) Géraldine Capodano (Methodology) Agnès Ducoulombier (Methodology) Julien Mancini (Methodology) Florence Duffaud (Methodology) Anthony Gonçalves (Methodology) Thémis Apostolidis (Conceptualization, Formal analysis, Methodology) Aurélien Proux (Conceptualization, Formal analysis, Methodology)

Funding

This study was partly supported by research funding from Institut National du Cancer, Région Provence-Alpes-Côte d’Azur and Canceropôle Provence-Alpes-Côte d’Azur in 2021 (Convention de financement N°2021-03).

Conflict of interest

The authors indicated no financial relationships.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.