Towards a novel approach guiding the decision-making process for anticancer treatment in patients with advanced cancer: framework for systemic anticancer treatment with palliative intent.

BACKGROUND: Weighing risks and benefits is currently the primary criterion for decisions regarding systemic anticancer treatment (SACT) in far advanced cancer patients, also in the modern immunotherapy- and molecular-targeted driven oncology. Decision aids rarely include substantially key concepts of early integrated palliative care (PC) and communication science. We compiled decisional factors (DFs) important for guiding the use of SACT with palliative intent (SACT-PI) and explored these DFs regarding their applicability in routine clinical care. PATIENTS AND METHODS: Clinician (participants: n = 28) and patient (n = 15) focus groups were conducted in an integrated oncology and PC setting. Thematic analysis was used to identify DFs. A Delphi survey of clinicians ranked the importance of DFs in routine decision-making. DFs were aligned with elements of the typical decision-making process, resulting in an eight-step guide for making SACT-PI decisions in clinical practice.
RESULTS: Eight focus groups revealed 55 DFs relating to established topics like providing information and risk-benefit analysis, as well as to PC topics like patients' attitudes, beliefs, and hopes; patient-physician interaction; and physician attitudes. Agreement on the relative importance was reached for 34 (62%) of 55 DFs, assigned to five elements: patient/family, clinicians/system, patient-clinician-interaction, information/patient education, risk-benefit weighting/actual decision. These themes are embedded in a potential clinically useful SACT-PI Decision Framework, which includes eight steps: assess, educate, verify, reflect, discuss, weigh, pause, and decide.
CONCLUSIONS: The SACT-PI Decision Framework integrates subjective patient factors, interpersonal factors, and PC issues into decision-making. Our findings complement existing decision aids and prompt lists by framing DFs in the context of SACT-PI and enforce the decision 'process', not the decision act. Further research is needed to explore the relative importance of DFs in specific patient situations and test structured decision-making processes, such as our SACT-PI Decision Framework, against standard care.

Introduction

Oncologists’ typically base treatment recommendations on survival, tumor control, and toxicity data from high-quality clinical trials. In patients with resistant tumors and in unfit patients, anticancer treatment is often applied, even without sufficient supporting evidence. Many patients may face toxicity with modest benefit from such treatments, whereas in others a systemic anticancer treatment with palliative intent (SACT-PI) may alleviate cancer-related symptoms.,

In modern oncology, patients with advanced disease often face uncertainty, and may have inaccurate perceptions of their prognosis, and treatment goals. Moreover, they may have difficulties weighing the toxicity risk against potential benefits in terms of symptom control and quality of life. These factors can compromise informed decision-making and may contribute to so-called aggressive end-of-life care. An individualized decision process becomes increasingly important to address palliative care (PC) domains like multidimensional symptom burden, illness and prognosis understanding, spirituality, life closure, or family involvement. Current evidence suggests that non-medical factors like attitudes, values and beliefs, sociodemographic variables such as (young) age and aspects of physician–patient relationship may influence SACT decisions. Current evidence to guide oncologists’ decision making for or against SACT-PI is rare.,

Decision support is a core component of early PC.,, Randomized, controlled trials of early specialized PC demonstrate improved quality of life for both patients and family members.21, 22, 23 In these trials, modified, less invasive palliative chemotherapy regimens were applied, resulting often in better quality of life and similar or even prolonged survival compared with standard regimens. Oncologists or other clinicians, supported by specialist PC clinicians, can deliver palliative interventions., Tools to improve oncologists’ alertness of symptoms, are reported to be effective for symptom management performance, whereas communication skills training improves communication, but not quality of decision. Existing decision aids (DAs) for advanced cancer patients are usually designed for specific cancer types, with a few being not disease-specific.29, 30, 31 They provide information on prognosis, treatment options including supportive care and adverse events. DAs apply rating scales to the possible harms and benefits of each option, elucidating patients’ preferences and include measures for value clarification. Such an approach, however, may not adequately touch upon essential subjective aspects, values, and burdens that are relevant when facing an advanced incurable illness and its treatments., In current DAs, such domains are infrequently covered, and if so, they are in a format similar to a prompt list.

Decision making is characterized as a process rather than a singular event, including information giving, eliciting patients’ values and preferences, weighing risks against benefits, and the actual decision.,, Reflecting personal values and knowledge about the natural course of the disease are essential to prepare for the decision-making encounter., Various clinicians [i.e. oncologists, general practitioners (GPs), nurses] typically involved in the care of patients also play an important role when patients face end-of-life decisions., Since current DAs in advanced incurable cancer do not incorporate non-medical factors and essential palliative domains in the decision process, we aimed to identify and characterize decisional factors (DFs) relevant to the decision process regarding SACT-PI in an interprofessional context applicable in routine clinical cancer care.

Methods

We used a mixed methods approach, beginning with focus groups (FGs) and ending in a Delphi survey. FGs were conducted with patients with advanced incurable cancer and clinicians from a tertiary cancer center (St. Gallen, Switzerland) to explore factors relevant for decisions on SACT-PI. FGs were selected in part because contributions amongst group participants may trigger the sharing of specific individual experiences, particularly in the context of PC. Using the decision factors generated from the FGs, we conducted a Delphi survey to determine the level of participant agreement on the relative importance of the factors. Previous studies have applied this combination of methods., The local ethics committee approved the study.

FG participants and sampling

Patients were approached for FG participation if they had a predefined disease situation in which further treatment was subject to considerable balancing by the oncologist (defined by the disease specialists of the cancer center for the specific cancer entities; Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2022.100496). They also had to have been involved in a decision-making process for or against palliative chemotherapy and confirmed cognitive ability to participate in the FG by their treating oncologist. Patients had to speak and understand German well. We used consecutive convenience sampling of all inpatients and outpatients and considered a balanced, diverse group, representing patients with and without actual SACT and main tumor types. A medical research fellow (NKM) from the section oncological palliative medicine recruited participants in the clinic in person or by phone. Patients received reimbursement of travel costs only.

For the clinician FGs, we used convenience sampling to assemble groups of experienced medical oncologists, GPs, and nurses at inpatient or outpatient oncology and palliative clinics.

Oncologists and senior clinical fellows from a tertiary cancer center (n = 40) and regional practicing general oncologists (n = 7) were randomly selected to be informed about the study and invited to participate. The final purposive sampling for two FGs assured the participation of both cancer center oncologist and practicing oncologist, and a diverse representation of oncology subspecialists.

Nurses involved in oncology or PC (n = 40) and a random selection of 10 out of 54 home care nurses from various regional services were invited. The final sampling included outpatient oncology, home care, PC inpatient and palliative consult service nurses.

From a publicly available database including 1000 GPs located in northeastern Switzerland, we selected by random sampling 110 who received a personal invitation to participate in the study. GPs were included if they cared for two or more patients with advanced cancer considering SACT-PI in the past 2 years. The final sampling assured the participation of younger and older, rural and urban, single and group practice GPs, and one or more GPs with reported dispute concerning SACT-PI.

Both patients and professionals provided written informed consent. Patients completed a Mini-Cog to screen for potential cognitive impairment. All participants completed a sociodemographic questionnaire, and specific questions, from either psychometric validated questionnaires or study-specific single questions (overview of assessments and original wording in German in supplement). Physicians and patients (not nurses) reported on their decision-making preferences using the Control Preference Scale, which is a single choice among five specific statements for patient-directed, shared, or oncologist-directed decision-making. Patients completed also the Edmonton Symptom Assessment Scale to assess their symptom burden, and they answered questions based on Ahn et al concerning their perceptions of their illness. Disease-related information was collected from medical records. Physicians and nurses were asked additional questions on their attitudes and practices regarding decision making [involvement of family in decision by oncologist, decision as event or process and perceived paternalism (oncologist only); involvement of nurse in decision-making (nurse only)], spiritual care, and their personal religion/spirituality. Physicians also reported their routinely applied PC interventions using five questions based on Jacobsen et al. Details on topics, measures, and response format for all questions are provided in the supplement (Supplementary Table S2, available at https://doi.org/10.1016/j.esmoop.2022.100496).

Running the FGs and analysis

A double-boarded oncologist/PC specialist (FS) and a psychologist (KR) conducted separate FGs for patients, physicians, and nurses. If patients preferred, they could choose an individual interview. The interview guideline (supplement) included an open, exploratory question on the topic and a range of factors considered relevant for decision-making in this phase of life. Questions were developed based on our own unpublished systematic literature review, and the professional experience of the interviewers. New themes generated in prior FGs were included in subsequent FGs.

FG discussions were audio-recorded and transcribed verbatim before thematic content analysis using atlas.ti.07. To establish the coding structure, two researchers (KR, NKM) independently coded a selection of transcripts. A deductive approach was used based on items from the interview guideline to identify topics and themes relevant to decision-making. In parallel, an inductive, data-driven approach was employed to identify additional topics and themes. The two coders discussed code names and definitions in-depth until consensus was reached. Subsequently all material was double-coded. The resulting list of codes was reduced by in-depth discussions among three researchers (KR, NKM, FS), applying the principles of paraphrasing, grouping, and integration and by verifying the new categories based on the original quotations.

Delphi survey and analysis

For the Delphi survey, codes were rephrased as statements, each addressing a factor considered to be important in decisions for/against SACT-PI. A professional translator was responsible for the English version of all statements. By the time the Delphi survey commenced, almost all patient participants had died. Because the intention of this step was to estimate the level of agreement on the importance of factors to be considered in SACT-PI decisions, rather than finding consensus, no new patients were recruited. Thus, the Delphi process only included clinicians who had participated in the FGs. In addition, five international experts in palliative cancer care were invited to respond to the Delphi survey. Participants completed the two-round Delphi survey online hosted on SurveyMonkey. In the first round, participants were asked to anonymously rate the importance of each statement for SACT-PI decision-making (scale: 0 = I do not agree; 10 = I fully agree). They could also comment on statements and suggest changes. The second round presented revised statements along with comments and minimal, median, and maximal ratings by physicians and nurses, respectively, so that respondents could re-rate each statement considering the group rating., The definition of a statement reaching ‘agreement’ was based on a statement meeting all three of the following criteria: (i) agreement mean value >7; (ii) a score ≥7 by 80% of respondents; and (iii) a score ≥7 rated by the international experts.

Integrating DFs in the decision-making process

The results of the Delphi survey made clear that it was necessary to incorporate the identified DFs relevant in the processes of clinical decision-making, although we did not anticipate this research question. Based on existing literature,,54, 55, 56 one author (NKM) proposed a conceptual model with five main elements of the decision-making process: (i) patient and family (predisposition, independent of current clinician); (ii) clinician and system (predisposition, independent of patient); (iii) clinician–patient interaction (individual patient with his/her individual clinician); (iv) information and patient education (including communicative and coordinative interventions delivered by the team); and (v) risk–benefit weighing and actual decision.,,, Three authors (NKM, KR, FS) independently assigned DFs to one of these five elements of the decision-making process. The three authors compared their assignments, discussed disagreements, and then re-assigned until they reached consensus.

SACT-PI decision framework

To integrate the retrieved DFs in the decision-making process of SACT-PI, we reviewed literature for systematic or systematized reviews on decision-making and implementation in oncology and for key concepts of integrated oncology and PC. The scoping literature review retrieved 102 quotations on decision making and 62 on implementation, and after further hand search and snowballing, 31 papers were consulted (citations: see Supplementary Material, available at https://doi.org/10.1016/j.esmoop.2022.100496). Models,57, 58, 59 and concepts characterizing collaborative multi-professional environments and communicative and self-reflective competences of involved professionals, a key PC domain, were mainly considered. The SACT-PI Decision Framework was generated using an iterative visualization process, continuously comparing whether results matched the model, until consensus was reached.

Results

Participant characteristics

Eight FGs and one private patient interview were conducted (Figure 1). Patients had various incurable cancer diseases diagnosed since 2 years (median), their perceived symptom burden was adequately controlled, and they had an ample illness and treatment understanding (Table 1). Sixty-seven percent (10/15) had active chemotherapy (trabectedin, carboplatin/paclitaxel, vinorelbine, oxaliplatin/gemcitabine, gemcitabine (n = 3), cabazitaxel, FOLFIRI, FOLFOX). Physicians were predominantly experienced oncology specialists and GPs who reported routinely applying PC interventions (Table 1). Nurses had professional experience of at least 2 years, worked in oncology and PC settings, and most reported a combination of the two in their current practice (Table 1).

Figure 1

Flow of study procedure.

Table 1

Characteristics of participants (patients and clinicians)

Patients (n = 15)
Age, median (min, max)
Years	66 (22, 82)
Sex, n (%)
Male/female	8/7 (53)
Education, n (%)
Basic/advanced	7/8 (47)
Living situation, n (%)
With partner or other persons/alone	14/1 (93)
Marital status, n (%)
Married/common law	12 (80)
Divorced/separated	2 (13)
Never married	1 (7)
Tumor type, n (%)
Gastrointestinal	7 (47)
Lung	3 (20)
Gynecologic and breast	2 (13)
Urological	1 (7)
Other	2 (13)
Current line of systemic anticancer therapy, n (%)
No systemic anticancer therapy	5 (33)
First line	2 (13)
Two or more lines	8 (53)
Goal of systemic anticancer therapy (multiple answers possible), n (%)
Symptom prevention	8 (53)
Symptom treatment	7 (47)
Life prolongation	6 (40)
Karnofski performance index, n (%)
>60 (able to care for most personal needs)	8 (53)
≤60 (unable to care for most personal needs)	6 (40)
Not specified	1 (7)
Course of events, median (min, max)
Time since cancer diagnosis (months)	24.6 (3, 56)
Time last chemotherapy to death (weeks, n = 11)	11 (3, 25)
Survival from focus group conduction (weeks)	14.3 (2, 40)
Patients’ perception of illness and treatmenta, median (min, max)
Awareness of diagnosis	4 (3, 4)
Awareness of prognosis	3 (2, 4)
Understanding of treatment options	3 (1, 4)
Goal of treatment is symptom control	4 (1, 4)
Goal of treatment is cure of disease	4 (1, 4)
Symptom control more important than cure	3 (1, 4)
Understanding of potential toxicities	4 (1, 4)
Health professional support for coping with disease	4 (3, 4)
Family support for coping with disease	4 (4, 4)
Thinking about end-of-life preparation	4 (2, 4)
Fight disease and thinking about end-of-life preparation	4 (2, 4)
End-of-life preparation important	4 (2, 4)
Decision-making preferencesb, n (%)
Doctor-directed decision	8 (53)
Shared decision	5 (33)
Patient-directed decision	1 (7)
Not specified	1 (7)
Symptom burdenc, median (min, max)
Tiredness	5 (1, 9)
General well-being	4 (1, 9)
Lack of energy	4 (1, 8)
Lack of appetite	3 (1, 10)
Breathlessness	2 (1, 7)
Pain	2 (1, 5)
Depression	2 (1, 4)
Anxiety	2 (1, 4)
Constipation	1 (1, 8)
Vomiting	1 (1, 8)
Nausea	1 (1, 4)

GPs, general practitioners.

Perceptions of illness and treatment (adapted from ).

Control Preference Scale ().

Edmonton Symptom Assessment Scale (scale range 1-10; higher scores indicate worse condition) ().

Family involvement by physician in decision: 1 = never, 2 = almost never, 3 = rarely, 4 = sometimes, 5 = often, 6 = almost always, 7 = always.

eInvolvement of nurse in decision: 1 = not at all, 2 = a little bit, 3 = quite, 4 = very much.

Paternalism defined as acting against inauthentic (patient’s) preferences 1 = never, 2 = rarely, 3 = often; 4 = very often (adapted from ).

Spiritual care: 1 = never, 2 = almost never, 3 = rarely, 4 = sometimes, 5 = often, 6 = almost always, 7 = always (adapted from ).

Religiosity and spirituality: 1 = not at all, 2 = a little bit, 3 = quite, 4 = very much (adapted from ).

Applied palliative care interventions: 1 = never, 2 = rarely, 3 = often; 4 = very often; (adapted from ).

One nurse did not answer this question.

DFs generated from FGs

The two independent coders identified 95 codes with a level of agreement on ∼75% of codes. For the identification of codes we did not use the results of the scoping literature review. These initial codes were eventually reduced to 55 codes and rephrased as statements representing DFs (Table 2 and Supplementary Table S3, available at https://doi.org/10.1016/j.esmoop.2022.100496). A substantial number of DFs related to psychosocial issues experienced by patients and their families encompassing patients’ hopes, attitudes, and beliefs, patients’ instrumental and emotional support, and family burden. Other factors addressed procedures of the health care system and specific attitudes of the clinicians. Many DFs related to the patient’s view of the clinician’s skills as a communicator, trust in the clinician, and characteristics specific to individual clinicians. Further DFs addressed information giving and verification of the patients’ understanding regarding treatment effects and the trajectory of illness. PC topics included end-of-life preparation and the opportunity to discuss sensitive topics. Finally, ‘classic’ DFs emerged such as weighing risks and benefits, defining goals, and the persons involved in the decision.

Table 2

Individual 55 factors with agreement on importance for a decision by Delphi survey

	Agreement on importance of factor based on Delphi survey	Agreement on assignment of factor to element of decision-making process	Decision on assignment to a decision process element by consensus	Arguments for decision on assignment
Clinician in agreement with patient decides in complex situation	a	2/3	Risk–benefit weighing	Actual decision important
Patient decides himself/herself	y	2/3	Risk–benefit weighing	Trustful relationship important
Offer to involve family members in decision	y	2/3	Information and patient education	Coordinative education, offer the team not oncologist
Offer to involve further specialists in decision	y	2/3	Information and patient education	Coordinative education, offer the team not oncologist
Ensuring good communication	y	3/3	Clinician–patient interaction
Existence of basis of trust	y	3/3	Clinician–patient interaction
Humanity/sympathy/honest interest of clinician	y	2/3	Clinician–patient interaction	Patient-perceived, not lone standing oncologist
Patient-perceived clinician’s competence	y	2/3	Clinician–patient interaction	Patient-perception of individual clinician
Patient-perceived clinician’s honesty	y	2/3	Clinician–patient interaction	Patient-perception of individual clinician
Patient-perceived clinician’s communicative abilities	y	2/3	Clinician–patient interaction	Patient-perception of individual clinician
Clinician’s knowledge of patient’s situation	y	3/3	Clinician–patient interaction
Patient’s feeling not being a stranger to clinician	a	2/3	Clinician–patient interaction	Dependent on individual patient and clinician
Willingness of patient and clinician to trust and go on the journey together	a	3/3	Clinician–patient interaction
Clinician’s intuition during decisional process	y	3/3	Clinicians and system
Inner conviction of clinician leads to unbalanced discussions	a	3/3	Clinicians and system
Being in good hands with health care team	y	2/3	Clinician–patient interaction	Not patient precondition unable to trust, but reality
Stability of patient’s relationship to clinician	y	3/3	Clinician–patient interaction
Offer to discuss existential topics with treatment team	y	3/3	Information and patient education
Practical burden on patient’s family	y	3/3	Risk–benefit weighing
Emotional burden on patient’s family	y	3/3	Patient and family
Support of patient by family	y	3/3	Patient and family
End-of-life preparations although CHT	y	3/3	Information and patient education
Understanding when inhibition of tumor growth	a	3/3	Information and patient education
Understanding of medical values	a	3/3	Information and patient education
Monitoring of patient understanding of information	y	3/3	Information and patient education
Monitoring of patient understanding of illness	y	3/3	Information and patient education
Patient overwhelmed by coping with illness	y	3/3	Patient and family
Exploring possible fears	y	3/3	Information and patient education
Planning time for consideration	y	2/3	Information and patient education	Information to patient of current standards, before the decision, not interactional
Neglecting end-of-life issues due to CHT activities	a	3/3	Clinicians and system
Meaning of hope for patient	y	3/3	Patient and family
Doing everything even though benefit unlikely	a	3/3	Patient and family
Hope connected with CHT	a	3/3	Patient and family
CHT same effect as preceding CHTs	a	3/3	Patient and family
CHT better effect if believed in it	a	3/3	Patient and family
CHT better effect if mental/emotional state good	a	3/3	Patient and family
General willingness to undergo CHT	a	3/3	Patient and family
Informational discussions prepared/planned	y	2/3	Information and patient education	Clinical standards of coordination
Amount of information during trajectory of illness	a	3/3	Information and patient education
Information on CHT administration/side-effects	y	3/3	Information and patient education
Information about treatment options	y	3/3	Information and patient education
Customized planning and logistics of CHT	a	3/3	Information and patient education
Regular evaluation of CHT effects and possibility of discontinuation	y	3/3	Information and patient education
Goals defined	y	2/3	Information and patient education	Goals are important before balancing
Weighing individual benefit versus cost	y	3/3	Risk–benefit weighing
Financial costs of CHT	a	3/3	Risk–benefit weighing	Financial toxicity
Meaning of prolonged survival time	a	2/3	Risk–benefit weighing	Balancing in the current situation expected OS benefit, with concrete goals
Use of remaining lifetime	a	3/3	Risk–benefit weighing	Concrete use in daily life
Impact of side-effects on QoL	y	3/3	Risk–benefit weighing
Impact of cancer on QoL	y	2/3	Risk–benefit weighing	Direct dependence on actual chemotherapy
CHT improves QoL	y	3/3	Risk–benefit weighing
CHT improves physical function	a	3/3	Risk–benefit weighing
CHT alleviates symptoms	y	3/3	Risk–benefit weighing
Health care team change without CHT	a	3/3	Clinicians and system
CHT for family	a	3/3	Patient and family

Shaded columns show assignment to elements of decision-making process.

a, ambiguous; CHT, chemotherapy; OS, overall survival; QoL, quality of life; y, yes.

Clinicians’ agreement on importance of DFs

Of the 28 clinicians from the FGs and the five international experts (total N = 33), 25 (76%) participated (14 physicians, 9 nurses, 2 experts) in the first round of the Delphi survey (Figure 1). In the second round, 20/25 (11 physicians, 7 nurses, 2 experts) participated. Agreement was reached for 34/55 (62%) DFs on their importance for decisions guiding SACT-PI (Table 2). Agreement was reached for most of the DFs relating to the clinician–patient relationship, patient education, and general information giving. In contrast, importance ratings were ambiguous for DFs related to patients’ hopes, attitudes, and beliefs, and to the meaning of remaining or prolonged lifetime.

Decision process elements and framework for SACT-PI

A total of 41 (78%) DFs were instantly assigned to one of the five decision process elements, whereas 14 required discussions within the research team until assignment, reflecting some overlap of the elements (Table 2). The elements are illustrated by exemplary quotations (Table 3).

Table 3

Elements of decision-making process: examples of quotations

Patient and family

Patients are overwhelmed by the situation; they must adapt first. This must not be underestimated. (Patient, C51)In the end, it’s your belief that determines whether the chemotherapy is effective or not. This is essential for me. If I want an improvement, I have to believe that it works. (Patient, C26)If I undergo chemotherapy, I believe in hope. (Patient, C7)If a patient experiences a lot of anxiety, particularly young persons, they do not want to die, and they grasp at straws, then I can understand the physician who doesn’t like to say it’s better to not give chemotherapy than to give palliative chemotherapy, but instead says we can try further chemotherapy. But sometimes I find it difficult when we know that it will not help any more, but we offered it to comfort the patient. (Nurse, C6)It should be considered from the beginning how we can support the patient’s family or address what the family members need, because sometimes we forget about them, and we require a lot from them. (Nurse, C29)

Clinicians and system

I try within the time I have available, sometimes I need 2-3 contacts, and sometimes it is enough to share the time walking along the hallway to sense certain things. (Physician, C45)A doctor that knows me will see that something is wrong with me, that my face is yellow. But if I go to a new doctor, he will not notice. (Patient, C42)The core business [of palliative care] is neglected due to chemotherapy and suddenly there are just three days left, and almost no time due to the patient’s health status to address existential issues or issues important to the patient. (Nurse, C52)

Clinician–patient-interaction

I had several oncologists, and I was surprised that each of them knew who I was, what I do, and what I prefer. I asked myself whether they had exchanged this information or how do they know it? I found it very agreeable, and I had full trust in each of them. (Patient C42)

Information and patient education

There are patients who want to know every detail—and they can cope with this information, while others cannot cope with too much detailed information; they are overwhelmed. (Nurse, C15)It is important to tell the patient that he/she can always stop chemotherapy […] or that we can try and see how he/she tolerates it, but that he/she can stop it anytime. (Physician, C21)The doctor could have told me more, how I will feel, how bad I will feel, and what I will have to struggle with. (Patient, C18)

Risk–benefit-weighing and actual decision

What is the current benefit for the patient, what is the range of side- effects he can expect, and how many good months are there in the end? Not only the months under treatment but the good months as well. If a patient has all the facts on the table, he/she can decide whether he/she will get on board and whether it is worth it. (Physician, C54)For me it is clear that […] without chemotherapy I have a better quality of life, even if it shortens my life. Therefore, I will live until then, try to be in a good mood and enjoy those things I still can do. (Patient, C5)The patient’s quality of life must be addressed, and what constitutes quality of life must be defined. What does quality of life mean to the patient? […] It is not a black box, but everybody has a different idea of what it means. The crucial criterion is what quality of life means to the individual patient. (Physician, C10)

As the first step towards the SACT-PI Decisional Framework, the decision process elements and individual factors were visualized by considering key elements of cancer PC and decisional models (references see Supplementary Material, available at https://doi.org/10.1016/j.esmoop.2022.100496). Assessment and communicative interventions are sentinel elements of integrated oncology and PC, therefore we linked: patients/family by ‘assessment’ to clinicians/system and coordinative/logistics interventions by ‘information and patient education’ to communicative interventions (Supplementary Figure S1, available at https://doi.org/10.1016/j.esmoop.2022.100496). Second, considering the importance of patient empowerment by education and information in PC, we combined them as steps (Supplementary Figure S2, available at https://doi.org/10.1016/j.esmoop.2022.100496).,, Likewise, appreciating the multi-professional team striving to esteem patient and families’ values by also reflecting own roles and biases, the steps verify and reflect were included.,, To appreciate the interrelation of discussion and weighing as a dynamic process, including also pauses, three steps were included., Third, the steps were aligned in a linear process including a balance (Supplementary Figure S3, available at https://doi.org/10.1016/j.esmoop.2022.100496). For the final SACT-PC Decisional Framework (Figure 2) the headings were specified. The interrelated themes ‘Assess patient and family’ and ‘Educate and inform patient’ were visually coupled as the following themes, ‘verify information and reinforce relationship’ and ‘reflect own role as clinician’. Our results underpin the importance of the four steps assess and educate, and verify and reflect. On this basis, the next four steps of the decisional process, which is often circular not linear,, are discuss, weigh, pause, and either decide or discuss again.

Figure 2

Systemic anticancer treatment with palliative intent (SACT-PI) decision framework.

Elements of decisional factors: subjects: (i) patient and family, (ii) clinician and system, (iii) interaction of clinicians and patients; actions: (iv) information and patient education, (v) risk-benefit weighting and actual decision.

Chemo, chemotherapy; GP, general practitioner, PS, perfotmance status.

Discussion

Traditionally, decisions regarding SACT-PI involve information provision and risk–benefit analysis, including the patient’s preferred involvement in decision-making. Our study identified additional DFs, which seem relevant for such challenging decisions, including various subjective patient factors, interpersonal factors, and PC issues. A key finding is the understanding that decisions regarding SACT-PI pertain to a process, not a single act, and that this process involves a sequence of interrelated coupled themes: first assess and educate, then verify and reflect, followed by the circular discuss-weigh-pause-decide process. These concrete steps are designed to guide the use of SACT-PI in an interprofessional clinical context and shall be considered as integral components of the decision-making process in the future.

DFs assigned to the process elements ‘information and patient education’ as well as ‘weighing risk–benefit and actual decision’ are covered in available, disease-specific DAs for advanced cancer patients., The explicit clarification of decision-making preference, usually an integral part of DAs, was not identified as a DF in our study. We hypothesize that the patient’s view of and involvement in decision-making may be a product of taking subjective and relational factors and PC domains into account, which were explicitly characterized in our study. Clarification of the individual meaning of cost and benefit of anticancer treatment played a prominent role in our results. These factors are less addressed in traditional DAs. Our findings also highlight that for SACT-PI, decision-making is not restricted to the patient and physician but the collaboration of various additional actors (family, multidisciplinary team).

DFs related to PC and end-of-life issues are covered partially in existing question prompt lists for patients with advanced cancer., These lists, however, are not customized to explicitly address PC domains in the context of a decision for or against chemotherapy, and patients tend to ask questions mainly from the prognosis section., Our data suggest that PC and end-of-life factors should play a more prominent role in the SACT-PI decision-making process.

We also identified subjective DFs that have been less addressed so far, such as those related to hope, attitudes, and beliefs., Conventional DAs aim to capture this subjective evidence by relatively simple ‘weigh scales’ to assist patients in becoming clearer about how much and why they value an option (so called ‘value clarification exercises’). Subjective evidence depends on mental representations of information, which may not be the verbatim information but rather the subjective interpretation of it. Such perceptions of reality shape decision-making, sometimes ‘outside of conscious awareness’. This mechanism may explain why those factors, resulting from a balancing process in the patient’s personal context, are scarcely applied in the typical medical context. Our DFs related to hope cover both a particular hope that aims at precise goals as well as a generalized hope implying absolute trust in the future without any specific predetermined goals. The clinicians in our sample were ambiguous regarding the importance of factors related to hope, attitudes, and beliefs for decision-making. This may reflect a general underestimation of subjective evidence, as previously reported.,

Quite a number of DFs cover the clinician–patient interaction and correspond with the conceptualization of trust in clinicians in general and oncologists specifically as overlapping dimensions of global trust, fidelity, competence, and honesty. Cancer patients have a strong need to trust in their oncologist and a deeper, more emotional bond of trust can be developed only after repeated interaction, two aspects that were clearly expressed by the patients during the FG discussions. Our results are consistent with the observation that in the advanced cancer setting, DAs and prompt lists may not facilitate decision-making without the sensitive endorsement by a clinician. Strikingly, some of our DFs related clearly to clinicians’ self-reflection and the role of intuition. This finding may be similar to the core attitudes described for professionals in PC, including honesty, perception, involvement, and intuition,, and is supported by a recent review that identified the predisposition of involved people and interactional aspects to be facilitators or barriers to shared decision-making. The SACT-PI Decision Framework seems applicable to modern oncology because the decision process is not based on prognostication, but on values, processes, and human interactions and may harmonize the often overrated hope linked to new anticancer treatments.

Limitations

This study was conducted at a European Society for Medical Oncology designated center of Integrated Oncology and PC. The SACT-PI Decision Framework may be different in other multidisciplinary cancer care settings. Our patients were all from the eastern, German-speaking part of Switzerland, a population with 25% foreigners, 50% having tertiary education and a low likelihood of poverty. Their view on DFs may not be representative for patients with a diverse cultural background and socioeconomic status. Decisional processes may be influenced by local cultures. For instance, certain topics such as spiritual care and faith did not emerge during our FG discussions despite prompts from the moderators. The preference for a ‘passive’ or ‘doctor-directed’ decision was high (53%, 8/15) compared with a median of 27% reported in studies conducted worldwide. The passive role was more frequent in Western (36%) compared with Asian (28%) countries, and in advanced (46%) versus early-stage (31%) cancer. The preference for a passive role in our patient population is consistent with systematic reviews in lung and colorectal cancer patients.,

In the context of the present Swiss health care system, nurses rarely have an explicit, independent role in their care for patients, which is quite different from other countries where both nursing and physician consultations on medical end-of-life decisions are common.,, In Switzerland, care provided by specialized, advanced practice nurses is only recently gaining traction. Therefore, the application of our SACT-PI Decision Framework to local clinics requires adaption to the roles and norms of the health care context.

Our findings may be enriched by more FGs with clinicians, even though we approached saturation, and by bigger patient FGs. In small groups participants are more likely to be involved emotionally. Also, the Delphi process, used as internal validation, may be expanded to patients.

Clinical implications and future research

This study deepens our understanding of factors relevant for decision-making in advanced, incurable cancer, a clinical setting where patient expectations about the effects of SACT-PI may be inaccurate, compromise decision-making, and lead to high cancer costs for a small benefit. Currently, the inclusion of patient/family and emotional issues is promoted as well as participative and supportive behavior in addition to traditional medical and technical issues. Likewise, early integration of palliative interventions is becoming standard of care,, and the role of specialized advanced practice nursing is emerging. The inclusion of a time out—to involve GPs—or a pause is a timely demand., The novel SACT-PI framework complements existing DAs and prompt lists by including themes related to the clinician–patient relationship, clinical benefit, or integrated oncology and PC and reinforcing the importance of the decision process, not the decision act.

Changing the practice of oncology clinicians can insufficiently be achieved by providing algorithms or guidelines, as shown by examples from chemotherapy-induced peripheral neuropathy or chemotherapy-induced nausea and vomiting. In contrast, training of specific skills can be effective in improving shared decisio-making. Another approach is to share responsibilities and interventions in the multidisciplinary and interprofessional care team. The success of specialist PC,,, or oncology nurses fostering resilience are examples.

The first two coupled SACT-PI steps ‘Assess patient and family’ and ‘Educate and inform patient’ can be implemented independent of the cultural or ethnical context; the steps include assessing patient preference for information and involvement in decision making. Clinicians other than physicians—namely oncology nurses—can carry out these two steps, or physicians working in a culture and clinical pathways of integrated oncology and cancer PC can be involved.

The second coupled SACT-PI steps require the oncologist who ‘verifies information and reinforces relationship’ and ‘reflects own role as clinician’. A conscious, consensually agreed and lived culture of interprofessional collaboration including transprofessional education that is institutionally supported and financed, may substantially improve the quality of decision processes. This includes strengthening the role of oncology nurses in decision-making processes.

For the implementation of the final four—circular—steps (discuss, weigh, pause, and decide) current DAs cover main elements. An additional important requirement would be to actively offer a pause as institutional standard of care,, and demand a training in shared decision-making skills.

Further research may investigate how to best operationalize such components into an applicable and useful clinical encounter, especially when decisions are difficult.

Conclusion

Decision making for anticancer regimens should include traditional risk–benefit considerations and standard information as delineated in available DAs. We also identified patient- and clinician-related factors, interpersonal relationships, and additional PC topics to be relevant for this treatment phase. These factors represent important complements to the standard practice of decision-making and require a step-by-step decisional process that is not sufficiently addressed in traditional DAs. Our proposal of a SACT-PI Decision Framework should be further investigated in order to determine how it can be implemented into the clinical treatment pathway and if it will improve patient, clinician, and system outcomes.