Shared Decision making aids in palliative care: A systematic review & meta-analysis of randomized controlled trials

Abstract

Background:

Shared Decision Making in palliative care encompassing patient-clinician collaboration has shown potential for improved patient outcomes. However, the impact of decision aids intended to support palliative care decision making discussions across clinical care settings remains uncertain.

Aim:

To assess the impact of shared of decision aids on decision making, clinical outcomes, and quality of life for patients with palliative care needs and their caregivers.

Design:

Systematic review and meta-analysis of randomized controlled trials, reported according to PRISMA, with overall evidence quality assessed using GRADE.

Data sources:

Ovid MEDLINE (all segments), EMBASE, Cochrane CENTRAL, Cochrane Database of Systematic Reviews, and Scopus, from inception through September 20, 2024, without restrictions on publication date, sample size, or language. Eligible studies were randomized controlled trials that included adult patients with palliative care needs or their caregivers, using any form of decision aids. Outcomes encompassed decision-related, clinical, and quality of life measures.

Results:

Sixteen randomized controlled trials were included. Participant ages ranged from 44 to 85 years. Predominant decision aids were videos or printed materials addressing goals of care and treatment options, with aims that included improving understanding of prognosis and supporting values clarification. Decision aids significantly improved decision-related knowledge (SMD: 0.40, 95% CI: 0.14–0.67, p = 0.003) and lowered preference for cardiopulmonary resuscitation (OR: 0.28, 95% CI: 0.20–0.40, p < 0.001). They showed no significant impact on decisional conflict, satisfaction, or regret.

Conclusion:

Evidence regarding the effectiveness of decision aids in palliative care decision making is limited and of low overall quality, with observed improvements primarily in patient knowledge and changes in stated treatment preferences rather than established improvements in decision quality or values-concordant care. Future research should prioritize integration into comprehensive shared decision making frameworks, cultural adaptations, and larger-scale studies to enhance patient-centered outcomes.

What is already known about the topic?

• Shared decision making can support patient engagement and knowledge in serious illness care.

• Decision aids are recognized as valuable tools to support decision making, but their impact in palliative care remains inconsistently demonstrated.

• Most studies assess decision aids as standalone tools, with limited integration into real-time clinical workflows.

What this paper adds?

• This meta-analysis shows that decision aids significantly improve patient knowledge and lower preference for cardiopulmonary resuscitation.

• Decision aids have no significant effect on decisional conflict, satisfaction, regret, anxiety, depression, values clarity, or quality of life.

• Tools embedded in clinician-facilitated conversations may be more likely to support preference-sensitive deliberation than standalone tools.

Implications for practice, theory, or policy?

• Decision aids alone are insufficient; they should be integrated into structured, clinician-facilitated decision making discussions.

• Policy efforts should support longitudinal, interactive models of shared decision making that reflect diverse cultural and relational decision making practices.

Introduction

Palliative care involves frequent, high-stakes, values-sensitive decisions^1,2—often under prognostic uncertainty and sometimes with surrogate involvement^3,4—where aligning treatment with what matters most to patients and families is central to quality care.^{5 –7} Shared decision making has been proposed as an approach to support this alignment.^{8 –15}

Shared decision making refers to a process of sharing information, deliberation, and decision making in a medium where patients and clinicians collaborate to identify care plans that are both medically sound and reflective of the patient’s preferences and values.^{8 –11} This approach enhances patient engagement, improves adherence, and supports informed treatment decisions.^{9,12 –15} Evidence indicates that shared decision making leads to better patient outcomes through discussions about options, risks, benefits, and personalized considerations combined with decision-support counseling. ¹⁶

Within the past two decades, decision aids have emerged as tools intended to strengthen shared decision making by presenting options in a balanced manner, clarifying potential outcomes, and—ideally—supporting values- and context-sensitive deliberation.^{11,16 –18} For example, a decision aid for cardiopulmonary resuscitation may compare attempting versus not attempting resuscitation, describe expected benefits and harms, and prompt patients to consider how each option aligns with their goals, whereas an informational handout may simply describe cardiopulmonary resuscitation without explicitly supporting choice deliberation. Decision aids are distinct from general educational materials or clinician counseling; they are designed to prepare patients (and caregivers, when applicable) to participate meaningfully in decision making. Yet in palliative care, decision aids may be used variably—sometimes as standalone materials rather than embedded within real-time clinical conversations—raising uncertainty about their effectiveness across patient-centered outcomes.^{2,19 –21}

The purpose of this systematic review and meta-analysis is to synthesize evidence from randomized controlled trials (RCTs) to evaluate the effectiveness of decision aids to manage patients with palliative care needs in any clinical care setting. The review aims to assess their impact on decision making, clinical outcomes, and quality of life measures for patients and their caregivers. Specifically, we evaluated whether decision aids, compared with usual care or alternative informational strategies, improve decision-related outcomes, clinical outcomes, and quality of life for patients with palliative care needs in any clinical care setting. By incorporating evidence from diverse populations, decision aid formats, and settings, the aim of this review is to provide clarity on the role of decision aids in supporting palliative care decision making discussions.

Methods

This meta-analysis followed an a priori protocol as well as reporting standards according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) ²² statement and Grading of Recommendations, Assessment, Development, and Evaluations guidelines (GRADE).^23,24 The review was designed to address the following question: among adults receiving palliative care (or their caregivers), what is the effect of decision aids, compared with usual care or alternative informational approaches, on decision-related outcomes, clinical outcomes, and quality-of-life measures? The protocol was developed prior to study initiation and guided eligibility criteria, outcomes of interest, and analytic methods. The protocol was not registered in PROSPERO; however, all methods were specified a priori and adhered to throughout the review process (Appendix 1).

Search strategy and study eligibility

A medical librarian collaborated with the study’s principal investigator to design and conduct a comprehensive search strategy. The search included several databases from their inception through December 20, 2024, without restrictions on publication date, sample size, or language. Controlled vocabulary terms were supplemented with relevant keywords to identify studies of decision aids or decision-support tools in palliative or serious illness care to support shared decision making. The search strategy included three core concepts: (1) serious illness/palliative or end-of-life care, including hospice, goals of care, and advance care planning; (2) decision aids or decision-support tools; and (3) randomized or controlled study designs, systematic reviews, or meta-analyses. Synonymous terms within each concept were combined with OR, and the three concepts were combined with AND. Databases searched included Ovid MEDLINE Epub Ahead of Print, Ovid Medline In-Process & Other Non-Indexed Citations, Ovid MEDLINE, Ovid EMBASE, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, and Scopus. The full search strategy for Ovid MEDLINE is provided in Appendix 2.

The review focused exclusively on RCTs. A preliminary scoping search identified sufficient randomized controlled trial evidence to justify restricting the review to RCTs, as this design allows the most rigorous evaluation of the comparative effectiveness and causal impact of decision aids on patient-centered outcomes while minimizing confounding. Eligibility criteria were predefined according to the Population, Intervention, Comparator, Outcome (PICO) framework and are summarized in Table 1.

Table 1.

Eligibility criteria for study inclusion.

Domain	Inclusion criteria	Exclusion criteria
Population	Adults (⩾18 years old) with advanced, serious, or life-limiting illness consistent with palliative care needs, or their caregivers/surrogate decision makers. Participants had to be facing a real healthcare decision related to treatment, goals of care, or advance planning.	Pediatric (<18 years old) participants; Participants without advanced or serious illness;
		Participants in simulation or hypothetical decision scenarios (e.g. standardized patients); and/or
		Participants not facing an actual healthcare decision.
Intervention	Any patient-related decision aids that were used to support shared decision making processes by presenting at least two healthcare options and their benefits/harms, with or without values clarification components, in any format (video, printed, electronic, or hybrid).	Educational materials not explicitly described as decision aids; clinician training programs without a patient-facing decision tool; interventions not involving structured comparison of care options; simulation tools or settings.
Comparator	Usual care without decision aids or with alternative informational strategies (e.g. standard consultation, educational materials), provided that the comparator did not include a structured shared decision making intervention equivalent to the decision aid.	Studies without a comparator group; comparators that included a structured shared decision making intervention comparable to the intervention arm.
Study design	Randomized controlled trials.	Studies without a comparator group; comparators, including a structured shared decision making intervention comparable to the intervention arm.
Outcomes	At least one of the following: decision-related outcomes (e.g. knowledge, decisional conflict, satisfaction, regret, preferences, values concordance), clinical outcomes (e.g. anxiety, depression, symptom burden), or quality-of-life measures.	Studies reporting only feasibility, acceptability, or implementation outcomes without reporting patient- or caregiver-level decision, clinical, or quality-of-life outcomes.
Setting	Real-world clinical settings (hospital, outpatient clinic, primary care, community, or nursing home).	Laboratory-based, vignette-based, or only-simulated based studies.
Language and date	No restrictions on language or publication date; databases searched from inception through December 31, 2024.	—

Eligible interventions were those explicitly designated as “decision aids” or “decision tools” in any format (e.g. paper, electronic, or hybrid). For the purposes of this review, decision aids were defined as structured tools or media intended to support deliberation by presenting at least two healthcare options and their potential benefits and harms, with or without explicit values clarification components, and designed to prepare patients or caregivers to participate in shared decision making. These decision aids aimed to facilitate decision making by supporting patient-clinician discussions, providing information on risks and benefits of treatment options, or incorporating patient values and preferences into care planning. During screening, we did not rely solely on the terminology used by study authors. Each intervention was assessed against our operational definition of a decision aid. Tools were considered decision aids if they supported an actual healthcare decision by presenting options and their benefits/harms and, when applicable, facilitating values clarification or deliberation. Materials that only provided general disease or treatment information without supporting option comparison or choice deliberation were considered informational tools rather than decision aids. Multicomponent interventions were eligible only when they included a patient- or caregiver-facing decision aid component.

Outcomes of interest

Outcomes of interest were prespecified a priori. Eligible studies were required to report at least one of the following outcome types: (1) decision-related outcomes (including, but not limited to, decision process; disease or treatment knowledge, understanding, or preference; decision-aid acceptance or understanding; decisional conflict; satisfaction with decision, consultation, or decision-aid; decision regret; preference for decision involvement or information; encounter time, barriers, and facilitators; clinician satisfaction); (2) clinical outcomes (including, but not limited to, worsened or improved symptoms; anxiety levels; depression; hospitalizations or emergency department visits); and/or (3) quality of life-related outcomes (use of specific quality of life scales).

Data screening and extraction

Data screening and extraction were conducted independently and in duplicate by two reviewers working blindly to each other’s assessments.

Data screening

• Abstracts retrieved through the search strategy were screened using standardized forms for inclusion and exclusion criteria. Reviewers were blinded to each other’s assessments during initial screening. Any conflicts were resolved through discussion, or, if necessary, by consultation with a senior investigator (AMAD). Studies meeting eligibility criteria underwent full-text review following the same process.

Data extraction

• For eligible studies, reviewers extracted data on patient demographics (e.g. sample size, age, race), baseline characteristics, study design, and relevant outcomes of interest. Discrepancies in extracted data were resolved through consensus or adjudication by a senior investigator (AMAD). In cases where data were missing or ambiguously reported, the original study authors were contacted directly to obtain clarification or additional information.

Risk of bias assessment and quality of evidence

The risk of bias or individual study quality was evaluated by two reviewers independently using the Cochrane risk of bias tool. ²⁵ Each article was rated as “high,” “unclear,” or “low” for performance, attrition, and reporting bias from each of six domains (random sequence generation; allocation concealment; blinding of participants and personnel; blinding of outcome assessment; incomplete outcome data, and selective reporting). The strength of evidence was evaluated using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach.^23,24

Statistical analysis

Comparisons of outcomes at follow-up between decision aid and control groups were made using a random-effects meta-analysis for continuous outcome variables. ²⁶ Because similar outcome domains (e.g. knowledge, decisional conflict, anxiety, quality of life) were assessed using different measurement instruments across studies (see Table 2), effect sizes were pooled using standardized mean differences (SMDs) and were estimated (using Hedges’ g) along with 95% confidence intervals (CIs) for each continuous outcome. Odds ratios (OR) with CIs were estimated for each binary outcome. Between-study heterogeneity was examined by estimating the I² statistic, which measures the proportion of variation between studies due to heterogeneity beyond chance. ²⁷ After applying a Bonferroni correction for multiple testing for the nine outcome measures, p-values <0.005 were considered as statistically significant. Comparators varied between usual care and active informational controls; however, data were insufficient to support formal subgroup analyses by comparator type. All statistical tests were two-sided. Statistical analyses were performed using R Statistical Software (version 4.1.2; R Foundation for Statistical Computing, Vienna, Austria).

Table 2.

Baseline characteristics and main findings of included randomized controlled trials.

First author’s last name, publication year (country)	Study objective	Intervention vs comparator (follow-up)	Design and setting	Clinician type		Decision aid type	Targeted population, N, mean age (intervention arm; control arm)	Primary outcome(s)	Main findings
				Person delivering or introducing intervention	Person discussing intervention			• All reported outcomes
Allen 2018 (USA)	To assess the effectiveness of a shared decision support intervention for patients considering destination therapy left ventricular assist device (DT-LVAD) placement.	Delivery of an in-person 2.5-h clinician-directed decision support training of local staff by the study team and formal integration of a 26-min video and 8-page pamphlet into each programs’ education process vs usual care 6 months)	Multisite, randomized, phased roll out (stepped wedge design) controlled trial in 6 mechanical circulatory support programs.	Primary physician	Primary physician	Support training, video, pamphlet	Patients considering DT-LVAD	Decision quality, measured by knowledge and values-choice concordance	Knowledge improved in the intervention group (difference-in-difference 5.5%, p = 0.03). Better alignment between patient values and stated treatment choice at 1 month (Kendall’s tau difference 0.28), not sustained at 6 months. No difference in decisional conflict, regret, quality of life, or depression.
							248 (63.2; 63.5)	• Knowledge/understanding (Author-developed study specific scale)
								• Decisional Conflict Scale
								• Decision Regret Scale
								• Treatment/Goals of Care preference
			In-person clinician-directed training, then material shown and discussed during formal DT-LVAD education consultation, and pamphlet taken home for self-use.					• Cardiopulmonary resuscitation/Intubation preference
								• Values (Author-developed study specific scale)
								• Treatment received at follow-up
								• Depression (Patient Health Questionnaire-2 score)
								• Health-related Quality of life measures (EuroQol scale)
Alsagheir 2020 (Saudi Arabia)	To evaluate the effect of decision aids for patients with metastatic colorectal cancer in the Arabic language.	Video that lists 2 treatment options (with or without chemotherapy) and presents the treatment benefit, response rates, and estimated survival rates vs usual care (6 months)	Multisite randomized controlled trial.	Research assistant	Primary physician	Video	Arabic speaking metastatic colorectal cancer patients	Patient’s understanding of prognosis and treatment information	No difference in knowledge between groups at baseline (4.16 vs 4.18 out of 8, p = 0.86) or at 1 month (4.85 vs 4.73, p = 0.45). Anxiety decreased over time in both groups, with no difference between intervention and control, but higher knowledge scores were associated with lower anxiety (estimate −0.53, p = 0.037).
			Video shown before consultation with the Primary physician at the outpatient clinic, then discussion with physician.				92 (55.3; 57)	• Knowledge/understanding (Author-developed study specific scale)
								• Anxiety and stress (General Anxiety Disorder 7 score)
El-jawahri 2010 (USA)	To determine whether the use of a goals-of-care video to supplement a verbal description can improve end-of-life decision making for patients with cancer.	Verbal description of possible levels of medical care followed by a 6-min video on the 3 levels of care vs verbal description of possible levels of medical care (immediate)	Single-site randomized controlled trial.	Research assistant	No discussion	Video	Returning outpatients with a diagnosis of malignant glioma	Preferences for care immediately after the intervention	Knowledge improved more in the video group than control (increase 1.9 vs 0.9 points out of 6, p = 0.004). Participants viewing the video were more likely to prefer comfort care (91.3% vs 22.2%, p < 0.0001) and less likely to prefer cardiopulmonary resuscitation after the intervention (8.7% vs 40.7%, p = 0.02). Lower decisional conflict in the video group (mean 13.7 vs 11.5, p = 0.002).
			After introductory discussion about the study with the oncologist, in the outpatient clinic. Verbal narrative read by Research assistant, then video shown.				50 (56; 51)	• Knowledge/understanding (Author-developed study specific scale)
								• Cardiopulmonary resuscitation/Intubation preference
								• Uncertainty (Decision Conflict Scale subscale)
El-jawahri 2015 (USA)	To examine the impact of a video decision tool for cardiopulmonary resuscitation and intubation on patients’ choices, knowledge, medical orders, and discussions with providers.	3-Min video on cardiopulmonary resuscitation and intubation vs regular inpatient care (12 months)	Two-site randomized controlled trial.	Research assistant	No discussion	Video	Inpatients over 60 yo with cancer or advanced illness and a prognosis of less than 1 year	Preferences for cardiopulmonary resuscitation and intubation immediately after the intervention	Knowledge higher in the video group (4.11 vs 2.45 out of 5, p < 0.0001). Lower preference for cardiopulmonary resuscitation (64% vs 32%, p < 0.0001) and intubation (72% vs 43%, p < 0.0001). More documented discussions with providers (81% vs 43%, p < 0.0001) and more orders to withhold cardiopulmonary resuscitation and intubation at discharge (57% vs 19% and 64% vs 19%, both p < 0.0001).
			Video shown at bedside with no interaction with Research assistant, on the inpatient internal medicine service.					• Knowledge/understanding (Author-developed study specific scale)
							150 (76; 76)	• Cardiopulmonary resuscitation/Intubation preference
								• Presence/Frequency of a discussion
								• Treatment received at follow-up
El-jawahri 2016 (USA)	To examine the impact of a video decision support tool and patient checklist on advance care planning for patients with heart failure.	Verbal description of the three possible goals of care followed by a 6-min video on the three levels of care and a Advance Care Planning checklist vs verbal description of the three possible goals of care (3 months)	Multisite randomized controlled trial.	Research assistant	Primary physician if patient decides	Video, Advance Care Planning Checklist	Ambulatory or Inpatients over 64 with advanced heart failure and limited prognosis	Preference for comfort care immediately after the intervention	Knowledge higher in the video group (4.1 vs 3.0 out of 6, p < 0.001). Greater preference for comfort-focused care (51% vs 30%, p < 0.001). Lower preference for cardiopulmonary resuscitation (32% vs 65%, p < 0.001) and intubation (24% vs 52%, p < 0.001). More goals-of-care discussions at 1 and 3 months (40% vs 6% and 61% vs 15%, both p < 0.001).
			During meeting with Research assistant but no interaction allowed, questions deferred to primary physician.				246 (81; 81)	• Knowledge/understanding (Author-developed study specific scale)
								• Cardiopulmonary resuscitation/Intubation preference
								• Presence/Frequency of a discussion
								• Treatment/Goals of Care preference
Eneanya 2020 (USA)	To test the efficacy of a video decision aid on knowledge of supportive kidney care among older patients with advanced chronic kidney disease.	Video education with an 11.5-min video on the different treatment options including hemodialysis, peritoneal dialysis and supportive kidney care vs verbal education with a short scriptbased on the script used for the video that was read aloud to each participant (immediate)	Multisite randomized controlled trial.	Research assistant	Not specified	Video, verbal script	Patients aged 65 or older with advanced chronic kidney disease	Knowledge of supportive kidney care	Knowledge of supportive kidney care increased after education in both groups (2.02–2.36 out of 3, p < 0.01) with no difference between video and verbal education (2.34 vs 2.38, p = 0.83). No difference in preference for supportive kidney care between groups before or after education (26% vs 30% vs 22%, p = 0.49). High satisfaction and acceptability reported in the video group.
			Verbal or video education were given to patients in a private room in the nephrology clinic, just before or after their clinic appointment.				100 (76; 75)	• Knowledge/understanding (Author-developed study specific scale)
								• DA acceptability and satisfaction (Author-developed study specific scale)
								• Treatment/Goals of Care preference
Hanson 2011 (USA)	To test whether a decision aid improves quality of decision making about feeding options in advanced dementia.	Surrogates received a printed decision aid with information about dementia, feeding options and the outcomes, advantages, and disadvantages of feeding tubes and assisted oral feeding vs usual care (9 months)	Multisite, cluster randomized controlled trial.	Research assistant	Primary healthcare provider if surrogate decides	Booklet or audio if impaired	Surrogates of nursing home residents with advanced dementia and feeding problems	Decisional Conflict Scale at 3 months	Knowledge higher in the intervention group (16.8 vs 15.1 out of 19, p < 0.001). Lower decisional conflict at 3 months (1.65 vs 1.97, p < 0.001). More discussions with healthcare providers (46% vs 33%, p = 0.04). No difference in decisional regret or satisfaction with decision.
			During meeting with Research assistant but no interaction allowed, then printed material taken home. Questions deferred to primary healthcare provider.				256 (59.3; 58.7)	• Knowledge/understanding (Author-developed study specific scale, and Expectation of Benefit Index)
								• Decisional Conflict Scale
								• Treatment/Goals of Care preference
								• Decision Regret Scale
								• Satisfaction with Decision/Decision Making (Satisfaction with Decision Scale)
								• Effective decision making (Decision Conflict Scale subscale)
								• Presence/Frequency of a discussion
								• Uncertainty (Decision Conflict Scale subscale)
								• Treatment received at follow-up
Leighl 2011 (Canada, Australia)	To evaluate the impact of a decision aid on patients with advanced colorectal cancer who were considering first-line systemic therapy	In-clinic material reviewing goals and types of treatments, including palliative care, and Take-home booklet with audio recording vs standard consultation (4 weeks)	Multisite, randomized controlled trial.	Primary physician	Primary physician	Booklet, audio	Outpatients of cancer clinics who had a diagnosis of incurable metastatic colorectal cancer and a low performance status	Understanding of prognostic and treatment information and satisfaction with decision making.	Knowledge improved more in the decision aid group between consultation and follow-up (increase 2.6 vs 0.8 points out of 16, p < 0.001). No difference in decisional conflict, satisfaction with decision, anxiety, or quality of life between groups. Treatment choices were similar.
			In the outpatient clinic, during initial consultation with oncologist regarding first-line systemic therapy.				207 (61; 62.5)	• Knowledge/understanding (Study specific scale adapted from a previous study)
								• Decisional Conflict Scale
								• Treatment/Goals of Care preference
								• Satisfaction with Decision/Decision Making (Satisfaction with Decision Scale)
								• Satisfaction with consultation (Study specific scale adapted from a previous study)
								• Preference for decision involvement
								• Anxiety and stress (Spielberger State-Trait Anxiety Inventory)
								• Health-related Quality of life measures (Functional Assessment of Cancer Therapy–General subscales)
Matlock 2014 (USA)	To assess the feasibility and acceptability of a decision aid designed for patients facing advanced or terminal illness.	A booklet and DVD entitled “Looking Ahead: Choices for medical care when you’re seriously ill” and a standard palliative care consultation vs standard palliative care consultation (5–20 days)	Single-site, pilot randomized controlled trial.	Research assistant	Palliative Care physician or Nurse if patient decides and has used the tool before or during the consultation	Booklet, video (DVD)	English-speaking patients from the inpatient palliative care consult service referred for goal clarification or end-of-life discussions; or their decision makers	Decision conflict, knowledge	No difference between groups in knowledge (62% vs 78% correct at follow-up, p = 0.35) or decisional conflict (15.8 vs 5.0, p = 0.41). Decision aid rated acceptable and empowering by participants.
			On the inpatient palliative care service, patients could review the decision aid at their convenience before, during or after the PC team consultation.				51 (55; 56)	• Knowledge/understanding (Author-developed study specific scale)
								• DA acceptability and satisfaction (Author-developed study specific scale)
								• Decisional Conflict Scale (10-item low literacy version)
Meropol 2013 (USA)	To test the hypothesis that the communication aid CONNECT would improve satisfaction with patient physician communication, reduce decisional conflict and generate more accurate expectations about the benefits and toxicities associated with treatment.	(a) Patient-directed online communication skills training (CONNECT) before consultation, with summary report to the physician vs (b) patient-directed online communication skills training (CONNECT) before consultation, without summary report to the physician vs standard consultation (3 months)	Multisite, randomized controlled trial.	Not applicable (online tool)	Primary physician	Online communication aid	Patients with a documented metastatic solid tumor scheduled for their first outpatient consultation with a medical oncologist	Satisfaction with patient-physician communication, decisional conflict scale, patients’ expectations	Greater satisfaction with communication in the intervention group, including discussion of quality of life (9.9 vs 9.0, p < 0.05) and format of communication (7.4 vs 7.0, p < 0.05). Participants reported that the program made decisions easier to reach (1.3 vs 0.9, p < 0.01) and increased satisfaction with decisions (1.6 vs 1.1, p < 0.001). No difference in decisional conflict overall.
			Patients self-use before their initial consultation with oncologist in the outpatient clinic.				743 ((a) 59.7; (b) 60; 58.8)	• DA acceptability and satisfaction (Author-developed study specific scale)
								• Decisional Conflict Scale Subscales
								• Satisfaction with Consultation (Author-developed study specific scale)
								• Uncertainty (Decisional Conflict Scale subscale)
								• Values Clarity (Decisional Conflict Scale subscale)
Mcilvennan 2018 (USA)	To characterize caregivers of patients considering destination therapy left ventricular assist device (DT-LVAD) and evaluate the effectiveness of a shared decision making intervention.	Delivery of an in-person 2.5-h clinician-directed decision support training of local staff by the study team and formal integration a the 26-min video and 8-page pamphlet into each programs’ education process vs usual care (6 months)	Multisite, randomized, phased roll out (stepped wedge design) controlled trial in 6 mechanical circulatory support programs across the United States.	Primary physician	Primary physician	Support training, video, pamphlet	Caregivers of patients considering DT-LVAD	Decision quality, measured by knowledge and values-choice concordance	Knowledge improved after education in both groups with no significant difference between intervention and control (adjusted difference 4.8%, p = 0.08). Stronger alignment between caregiver values and stated treatment choice at 1 month in the intervention group (Kendall’s tau difference 0.36, p = 0.03), not associated with actual treatment received at 6 months. Decisional conflict decreased less immediately after education in the intervention group (p = 0.009), with no difference at 1 or 6 months. No difference in decision regret, depression, stress, preparedness, or satisfaction with decision.
			Used an in-person clinician-directed training, then material shown and discussed during formal LVAD education consultation, and pamphlet taken home for self-use.				182 (62.2; 60.2)	• Knowledge/understanding (Author-developed study specific scale)
								• DA acceptability and satisfaction (Author-developed study specific scale)
								• Decisional Conflict Scale
								• Treatment/Goals of Care preference
								• Decision Regret Scale
								• Satisfaction with Decision/Decision Making (Family Satisfaction With Care in the Intensive Care Unit-24 subscale)
								• Values (Author-developed study specific scale)
								• Treatment Received at Follow up
								• Anxiety and stress (Perceived Stress Scale)
								• Depression (Patient Health Questionnaire-2)
Oostendorp 2017 (Netherlands)	8 weeks	Standard oncology consultation followed by a nurse consultation introducing the two second-line options (best supportive care or palliative chemotherapy), with oral explanation by the nurse of the types of information available in the printed decision aid. The patient chose which information to receive and received a personalized booklet to take home vs standard oncology consultation (8 weeks)	Multisite randomized phase II controlled trial.	Nurse	No discussion	Verbal information during consultation, Booklet	Previously contacted patients with advanced breast or colorectal cancer who were randomized when they experienced disease progression and were offered second-line chemotherapy	Anxiety at 1 week and 8 weeks.	No difference in anxiety, depression, quality of life, or cancer worries between groups at 1 or 8 weeks. Higher subjective knowledge in the intervention group (6.7 vs 6.3, p = 0.022) and stronger treatment preference (3.0 vs 2.5, p = 0.030). No difference in objective knowledge, risk perception, decision satisfaction, or treatment received.
			During outpatient consultation with a nurse, 1 week after consultation with oncologist, then self-use of booklet at home.				128 (61; 62.5)	• Knowledge/understanding (Author-developed study specific scale)
								• Treatment/Goals of Care preference
								• Uncertainty (Decisional Conflict Scale subscale)
								• Anxiety and stress (Hospital Anxiety and Depression Scale)
								• Depression (Hospital Anxiety and Depression Scale)
								• Health-related Quality of life measures (European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 15 Palliative Care)
Shirai 2012 (Japan)	To evaluate the patients’ perception of the usefulness of a question prompt sheet in facilitating the involvement of advanced cancer patients during consultation.	Structured list of questions covering the items a patient may want to ask their physicians regarding their illness and treatment (Question Prompt Sheet ) and Hospital Introduction Sheet vs hospital Introduction Sheet (immediate)	Single-site, randomized controlled trial.	Research assistant	Primary physician	Question Prompt Sheet	Patients with advanced cancer presenting for their first consultation at thoracic oncology or gastrointestinal oncology division	Patient rating of the usefulness of the material	Higher perceived usefulness of the material for asking questions (4.4 vs 2.7 out of 10, p = 0.033) and for future consultations (5.3 vs 2.8, p = 0.006). No difference in satisfaction with consultation, number of questions asked, or understanding of the treatment plan.
			After first oncology consultation in the outpatient clinic, patients received the decision aid for self-use at home before their next consultation.				63 (63.5 (Median); 64 (Median))	• DA acceptability and satisfaction (Questionnaire adapted from a previous study)
								• Satisfaction with Consultation (Questionnaire adapted from a previous study)
Stegmann 2020 (Netherlands)	To assess the utility of the Outcome Prioritization Tool, designed to aid discussion with a patient in regards to their treatment goals, to empower patients with cancer through structured conversations about generic treatment goals with general practitioners.	Discussion with General Practitioner using the Outcome Prioritization Tool decision-aid before oncology consultation vs usual care (immediate)	Multisite randomized controlled trial.	General practitioner	General practitioner	Outcome Prioritization toolprinted scales	Patients 60 years or older, with cancer that had no curative treatment options, and had to make a treatment decision with an oncologist	Patient empowerment, as measured by the decision self-efficacy scale.	No difference in patient empowerment between groups (86.8 vs 84.2, p = 0.47). Lower anxiety in the intervention group (6.0 vs 7.6, p < 0.05) and less mild fatigue (58.8% vs 77.2%, p = 0.05). No difference in depression.
			Patients received the DA during a consultation with their general practitioner before consultation with the oncologist in the outpatient clinic.				114 (75.3; 72.9)	• Decision self-efficacy scale
								• DA acceptability and satisfaction (Questionnaire adapted from a previous study)
								• Anxiety and stress (Hospital Anxiety and Depression Scale)
								• Depression (Hospital Anxiety and Depression Scale)
								• Health-related Quality of life measures (Multidimensional Fatigue Inventory)
Volandes 2013 (USA)	To examine the effect of a video decision support tool on cardiopulmonary resuscitation preferences among patients with advanced cancer.	3-Min video on cardiopulmonary resuscitation and intubation, and verbal description of cardiopulmonary resuscitation vs verbal description of cardiopulmonary resuscitation (8 weeks)	Multisite randomized controlled trial.	Research assistant	No discussion	Video	Returning outpatients with advanced cancer who were aware that the prognosis was less than 1 year	Preference for cardiopulmonary resuscitation immediately after the intervention	Knowledge higher in the video group (3.3 vs 2.6 out of 4, p < 0.001). Lower preference for cardiopulmonary resuscitation after the intervention (20% vs 48%, p < 0.001), with a similar trend at 6–8 weeks.
			In the outpatient clinic, directly after consultation with oncologist.				150 (63; 62)	• Knowledge/understanding (Author-developed study specific scale)
								• DA acceptability and satisfaction (Author-developed study specific scale)
								• Cardiopulmonary resuscitation/Intubation preference
Yun 2011 (Korea)	To test whether a decision aid explaining how to discuss the approach of death with a family member with cancer would help family caregivers decide to discuss a terminal prognosis.	20-Min take-home DVD + companion 43-page workbook (“Patients want to know the truth”) vs 20-min take-home DVD on pain management (“Controlling Cancer Pain: A Video for Patients and Families”) and 29-page educational book on pain control (“Cancer Pain Can Be Controlled”; 6 months)	Multisite randomized controlled trial.	Research assistant	No discussion	Video, workbook	Primary caregivers of patients who were diagnosed with terminal cancer and who were capable of being aware of a terminal condition	Decision of the caregiver to discuss a terminal prognosis with the patient after 1 month	>Lower decisional conflict in the intervention group at 1 month, including conflict (p = 0.003), uncertainty (p = 0.019), values clarity (p = 0.007), and total decisional conflict score (p = 0.008), with effects persisting over 6 months. Lower depression scores in the intervention group from baseline to 1 month (p = 0.007) and sustained over 6 months (p = 0.008). No difference in satisfaction with decision, anxiety, quality of life, or decision regret.
			Caregivers received the DA for self-use at home.				444 (46.6; 44.6)	• Decisional Conflict Scale
								• Decision Regret Scale
								•Effective decision making (Decisional Conflict Scale subscale)
								• Uncertainty (Decisional Conflict Scale subscale)
								• Values Clarity (Decisional Conflict Scale subscale)
								• Anxiety and stress (Hospital Anxiety and Depression Scale)
								• Depression (Hospital Anxiety and Depression Scale)
								• Health-related Quality of life measures (Caregiver Quality of Life – Cancer)

Results

Systematic review findings

The initial search strategy yielded 3561 citations, of which 16 RCTs^{28 –43} were deemed to be eligible and included in the final analysis as shown in the study search and selection flowchart (Figure 1). The number of randomized participants in the included studies ranged from 51 to 743, with a patient or surrogate age ranging from 44 to 85 years. Follow-up periods varied between same day and 12 months post-intervention. The studies encompassed diverse settings and geographic locations: of the 16 included studies, there were 10 trials conducted in the USA^{29 –33,35 –37,40,42}; two in the Netherlands^38,43; one study in Japan ³⁹ ; one in Korea ⁴¹ ; one in Saudi Arabia ²⁸ ; and one multisite study in Australia and Canada. ³⁴

Figure 1.

Flowchart of the study search and eligibility.

Of the 16 studies, 12 enrolled patients exclusively^{28 –32,34,37 –40,42,43}; one enrolled both patients and caregivers/surrogates ³⁵ ; and three enrolled only caregivers or surrogates.^33,36,41 Most decision aids used videos and/or printed materials.^{28 –35,38 –41,43} Four studies^34,36,42,43 actively supported shared decision making by facilitating bidirectional patient-clinician communication. The remaining studies provided decision aids for independent review, without structured shared decision making documentation. The included interventions were grouped into three broad typologies: (1) standalone patient- or caregiver-facing decision aids, most commonly videos and/or printed materials presenting options and benefits/harms; (2) clinician-facilitated decision aids embedded within structured shared decision making encounters, including multicomponent interventions that combined clinician training with patient-facing video or pamphlet tools; and (3) digital or online communication aids designed to prepare patients for consultation.

For decision aids formats, 13 studies used videos and/or printed materials (e.g. booklets)^{28 –35,38 –41,43}; two studies combined physician training with a video and pamphlet decision aid^36,42; and one study directed patients to an online communication aid, ³⁷ as shown in Table 2. Comparators included usual care, verbal information alone, or standard clinician-led discussions without structured decision-support tools. Most study interventions were conducted in hospital or clinic settings and involved the presence of a clinician.

Reported decision making related outcomes included knowledge and understanding, treatment preferences, and decisional conflict. Clinical outcomes, such as anxiety, stress, and depression, were also frequently reported. Quality-of-life measures, though less common, were also assessed.

Baseline characteristics and main findings of the studies are detailed in Table 2. Risk of bias was generally moderate (Appendix 3), and overall evidence quality was graded as low, mainly due to indirectness and inconsistency.

Meta-analysis

Meta-analyses were conducted for nine outcomes across the 16 included studies as shown in Table 3. Meta-analysis results for individual outcomes are summarized in Figure 2 and Appendix 4. Figure 2 presents pooled effects for decision-related outcomes, including knowledge, decisional conflict, satisfaction, regret, and values clarity. Figure 3 (Appendix 4) presents pooled effects for clinical and quality-of-life outcomes, including anxiety, depression, quality of life, and stated cardiopulmonary resuscitation preference. Across outcomes, effect sizes were estimated using standardized mean differences or odds ratios, with heterogeneity varying substantially between outcome domains (I² range: 0%–85%).

Table 3.

Between-group differences in outcomes.

Outcome	Number of studies	Decision aid sample size	Controls sample size	Standardized mean difference − decision aid minus controls (95% CI)	p-Value	I2 (%)
Knowledge	12	823	874	0.40 (0.14, 0.67)	0.003	85
Decisional conflict	5	535	599	−0.07 (−0.33, 0.19)	0.603	78
Satisfaction with decision	4	346	361	0.05 (−0.20, 0.31)	0.693	65
Decisional regret	4	306	359	0.10 (−0.15, 0.34)	0.442	61
Quality of life	5	460	460	−0.06 (−0.19, 0.07)	0.371	0
Depression	5	406	449	−0.08 (−0.27, 0.11)	0.416	47
Values	4	555	503	0.01 (−0.28, 0.30)	0.952	82
Anxiety	3	326	359	−0.07 (−0.22, 0.09)	0.398	14
Outcome	Number of studies	Decision aid sample size	Controls sample size	Odds ratio decision aid vs controls (95% CI)	p-Value	I2 (%)
Cardiopulmonary resuscitation treatment preference	4	291	305	0.28 (0.20, 0.40)	< 0.001	0

CI: confidence interval.

p-Values result from random effects models.

Figure 2.

(a) Forest plot for difference in knowledge between decision aid and control groups. Knowledge assessed using study-specific or adapted knowledge questionnaires with varying score ranges; pooled using standardized mean differences. (b) Forest plot for difference in decisional conflict between decision aid and control groups. Decisional conflict assessed using the Decisional Conflict Scale and related subscales; pooled as standardized mean differences. (c) Forest plot for difference in satisfaction with decision between decision aid and control groups. Satisfaction with decision assessed using Satisfaction With Decision Scale or study-specific measures; pooled as standardized mean differences. (d) Forest plot for difference in decisional regret between decision aid and control groups. Decisional regret assessed using the Decision Regret Scale; pooled as standardized mean differences. (e) Forest plot for difference in cardiopulmonary resuscitation treatment preference between decision aid and control groups. Preference for cardiopulmonary resuscitation assessed as a binary outcome; pooled using odds ratios.

Decision aids were associated with significantly higher knowledge scores (SMD = 0.40; 95% CI: 0.14–0.67; p = 0.003) and changes in stated treatment preferences, including lower preference for cardiopulmonary resuscitation as part of patients’ stated goals of care (OR = 0.28; 95% CI: 0.20–0.40; p < 0.001). For all other outcomes—including decisional conflict, satisfaction with decision, decisional regret, quality of life, anxiety, depression, and values clarity—there were no statistically significant differences between intervention and control groups. Heterogeneity varied across outcomes, with I² ranging from 0% to 85%.

Discussion

Main findings

This systematic review and meta-analysis evaluated the impact of decision aids intended to support palliative care decision making discussions within clinical care settings. Findings suggest that while decision aids significantly improve patient knowledge and reduce stated cardiopulmonary resuscitation preference, they have limited impact on other decision-related outcomes, such as decisional conflict, satisfaction, or regret. The quality of evidence was rated generally low due to indirectness (i.e. reflecting variability across studies in patient populations, decision aid formats, clinical contexts, outcome measures, and the extent to which interventions reflected real-world shared decision making processes) and inconsistency (i.e. reflecting heterogeneity in effect estimates across trials).

These findings reinforce prior evidence that decision aids—while effective in improving knowledge—are only one component of shared decision making, and may be insufficient on their own to address the complex, emotional, and relational dimensions of serious illness decision making.^{11,16,17,43,44} In this review, decision aids were associated with significant improvements in knowledge and reduced preference for cardiopulmonary resuscitation, yet no consistent effects were observed for decisional conflict, satisfaction, or regret—outcomes that are more closely tied to values clarification, clinician–patient interaction, and emotional support.

For example, visual tools depicting cardiopulmonary resuscitation have been associated with changes in stated treatment preferences, including lower preference for cardiopulmonary resuscitation, likely by improving understanding of the intervention, its likely outcomes, and its potential burdens; However, such tools must be designed and implemented carefully to minimize framing effects and support voluntary, informed decision making rather than directing patients toward a particular choice.^31,40 This improved understanding supports informed consent and facilitates deeper discussions about prognosis, trade-offs, and treatment goals.^1,2,29 However, most included trials deployed decision aids as standalone interventions, with only a minority explicitly integrating them into clinician-facilitated shared decision making processes, which may help explain the limited impact observed on relational and affective outcomes. Rather than limiting shared decision making use to situations of clinical equipoise, the emphasis should be on how well these tools are implemented to support authentic patient engagement across care settings. ⁴⁵

The central role of communication

Knowledge alone does not ensure values-congruent choices. Decisional conflict, regret, and satisfaction are shaped by relational and emotional factors—elements that static tools may not fully address.^8,11,19,46 Tools that were not embedded in bidirectional, clinician-facilitated workflows showed minimal impact on these outcomes. In contrast, studies that combined decision aids with structured communication—such as clinician-guided values elicitation, shared agenda setting, iterative dialogue, or decision coaching—demonstrated better alignment with care preferences and greater patient satisfaction.^34,36 These interventions allowed clinicians to contextualize information, address emotional concerns, and help patients deliberate in accordance with their values. Thus, decision aids should be integrated into shared decision making conversations—not used in isolation.^9,10 Yet, despite professional advocacy, shared decision making is often inconsistently implemented as part of palliative care discussions due to systemic barriers and clinician discomfort with prognosis discussions, which can undercut the intended benefits of decision aids. ⁴⁷

Importantly, decision aids did not increase anxiety or depression, supporting their emotional safety even in sensitive contexts.^28,41 Still, reducing psychological distress likely requires adjunctive support, such as coaching or structured debriefs.^48,49 Embedding decision aids into ongoing relational conversations, rather than one-off interventions, may provide the continuity and support patients need to process complex emotions over time. ⁵⁰ The lack of improvement in values clarity points to gaps in either decision aid design or how these tools are implemented. Tools aligned with International Patient Decision Aid Standards (IPDAS) ⁵¹ that incorporate values clarification are more likely to support patient-centered decisions.^16,48 Clinicians should fill these gaps by explicitly inviting patients to reflect on their hopes, fears, and factors that matter most. Without guided reflection or follow-up, patients may struggle to connect the information presented in decision aids to their own priorities. Thus, well-designed tools may fall short unless they are embedded within supportive communication that actively engages patients in clarifying their values.

Implementation gaps and the need for cultural tailoring

Most included studies deployed decision aids for independent use with minimal follow-up. Only a few incorporated meaningful, two-way communication. This likely explains the limited effect on broader outcomes like decisional conflict or satisfaction.^9,10 Training clinicians to respond effectively to what patients learn from decision aids—and to deepen the conversation beyond factual comprehension—is essential. This includes developing professional competencies in recognizing and responding to patient cues, using structured follow-up questions (e.g. “What concerns you most about this decision?”), and incorporating patient-reported goals and values into the care plan. ⁴⁹ Decision aids must be complemented by deliberate clinician behaviors—such as eliciting goals, exploring trade-offs, and clarifying values—if they are to fulfill their role in shared decision making.^45,47

Lastly, decision aids, particularly in palliative care discussions, must reflect the social and cultural realities of care.^10,14,16,44 Most trials were based in high-income, Western contexts and assumed individual decision making models. Yet many patients— especially in collectivist cultures— make decisions in partnership with family or spiritual advisors. Equity-oriented shared decision making approaches should prioritize cultural responsiveness and ensure tools are co-designed with communities to avoid perpetuating structural inequities. ⁵² Cultural tailoring may include involving family caregivers in the design process, offering multilingual formats, and framing decisions in ways that respect spiritual and cultural beliefs/values. Future tools should be designed for and tested in diverse populations and real-world care settings.^21,44 Technologies like patient-reported outcome dashboards may help identify evolving symptoms and priorities to guide timely, values-aligned decisions ⁴⁹ —especially in the face of declining decision making capacity.^6,7,49

Limitations and future directions

This review is limited by substantial heterogeneity in study designs, outcome measures, intervention characteristics, and participant demographics, which complicate cross-study comparisons and limit the precision of pooled estimates. Decision aids varied widely in format, content, and degree of integration into clinical encounters, and similar outcome domains (e.g. knowledge, decisional conflict, quality of life) were assessed using different measurement instruments across trials. Most included studies were conducted in high-income settings with limited cultural diversity, reducing generalizability and transferability to more diverse populations and care models. Additionally, many studies provided only short-term follow-up, limiting insights into long-term outcomes such as patient empowerment, caregiver burden, downstream care utilization, and sustained alignment of care with patient values. The overall low quality of evidence, influenced by indirectness and inconsistency—reflecting variability in populations, interventions, implementation contexts, and heterogeneity in effect estimates—further constrains the strength and transferability of results.

Nonetheless, this review is strengthened by an a priori protocol, comprehensive search strategy, and quantitative synthesis across a range of decision-related outcomes with methodologically rigorous rating of evidence.^24,25 The results remain valuable in illustrating current gaps and offering a roadmap for improving decision aid implementation in palliative care. Future research should focus on developing and evaluating decision aids specifically tailored to the nuanced demands of palliative care. This includes designing culturally sensitive decision aids that acknowledge family involvement in decision making, training clinicians to facilitate emotionally supportive dialogues, and incorporating patient values and patient-reported outcome data to guide real-time decisions, while acknowledging the diversity of shared decision making approaches that can be leveraged.^53,54 Long-term studies and mixed-methods approaches through validated frameworks should be employed to understand both the quantitative impacts and the qualitative experiences of patients, caregivers, and clinicians using decision aids in palliative care.^{55 –59} Ultimately, improving decision aid implementation and integration into shared decision making frameworks may support patient- and family-centered care that respects individual values, goals, and the lived context of serious illness and end-of-life care.^2,44,46

Conclusions

This systematic review highlights limited and generally low-quality evidence regarding the impact of including decision aids in palliative care discussions in diverse clinical settings. Decision aids were associated with improved knowledge and changes in stated treatment preferences, including lower preference for cardiopulmonary resuscitation, but these findings should not be interpreted as evidence of superior decision quality or values-concordant care. Their standalone use may be insufficient to support the emotional, relational, and contextual work of serious illness decision making. Future research should focus on co-designed, culturally tailored decision aids that minimize framing effects, support voluntary informed decision making, and are embedded within clinician-facilitated shared decision making workflows. Mixed-methods and real-world studies are needed to examine how decision aids shape understanding, deliberation, values clarification, patient experience, and goal-concordant care for patients and their care partners.

Appendix 1

Protocol: Decision aids for shared decision making in palliative care

Background

Palliative care involves complex chronic medical care that encompasses high-load of information presenting advanced-care life decision, treatment plans, medication choice, advanced directives, and side-effects management. In addition to that, this process often accompanied with emotional, financial and social dimensions. Because patients are faced with these various challenges, the presence of decision aids that may guide them to make an informed and shared decision are valuable and helpful in shouldering all challenges they face while continuing to live their lives with the best quality possible.

Objective

We aim in this review to synthesize the evidence about the use of decision aids on different outcomes for patients with palliative care needs and/or their caregivers.

PICO 1

Population

Patients with palliative care needs using decision making aids or tools in any clinical setting (no simulation). We will include all patients age; in cases where the patient may not have agency to make an informed decision (e.g. young pediatrics/elderly with advanced dementia), the caregiver will be considered the decision-maker. Patients in eligible studies will need to be facing an actual decision of their care and lives.

Intervention

We will include randomized controlled trials (RCTs) or systematic reviews and meta-analyses that included RCTs that evaluate any intervention aiming to improve decision making between clinicians and patients or their caregivers. We will include any decision making intervention type that aimed on informing patient decisions or focused on informed consent, reported information exchange between clinicians and patients encompassing the discussion of risks and benefits of treatment options as well as values and preferences.

Comparator

Usual care or any modified comparable intervention

Outcomes (any of the following):

• Decision-related outcomes: Decision process, satisfaction, participation, attitude, regret, preferences, encounter time, barriers and facilitators, clinician satisfaction.

• Clinical outcomes (disease-specific), quality of life, worsened or improved symptoms,

• Capacity/workload outcomes: burden of treatment and illness, resource availability or access to services

Appendix 2

Ovid

Database(s): APA PsycInfo 1806 to December, 2024 Week 3, EBM Reviews - Cochrane Central Register of Controlled Trials December 2024, EBM Reviews - Cochrane Database of Systematic Reviews 2005 to December 31, 2024, Embase 1974 to 2024 December 31, Ovid MEDLINE(R) and Epub Ahead of Print, In-Process, In-Data-Review & Other Non-Indexed Citations, Daily and Versions 1946 to December 31, 2024

Search Strategy:

#	Searches	Results
1	exp Terminal Care/	161,407
2	exp Critical Illness/	92,106
3	exp Hospices/	26,374
4	(“critical illness*” or “critically ill” or “end of life” or “goals of care” or hospice* or “life-limiting” or “serious illness*” or “seriously ill” or terminal or “terminally ill”).ti,ab,kf.	1,474,087
5	1 or 2 or 3 or 4	1,579,421
6	exp palliative therapy/	238,668
7	exp Palliative Care/	258,601
8	((symptomatic adj1 (treatment* or therap* or care or surg* or manag* or intervention* or radiotherap*)) or palliation or palliative).mp.	438,333
9	6 or 7 or 8	440,537
10	exp Advance Directives/	21,895
11	exp living will/	15,357
12	(((advance or advanced) adj3 directive*) or ((medical or healthcare or “health care” or durable) adj2 (power adj2 attorney)) or “advance care” or “advanced care” or “living will” or “living wills” or “psychiatric will” or “psychiatric wills” or “ulysses contract*”).mp.	53,469
13	10 or 11 or 12	53,469
14	5 or 9 or 13	1,925,928
15	*Decision Making, Computer-Assisted/	15,647
16	*Decision Support Techniques/	28,008
17	*Decision Support Systems, Clinical/	13,802
18	*decision support system/	17,693
19	*patient decision making/	2955
20	(((decision* or decid*) adj2 (support* or aid* or tool* or instrument* or technolog* or technique* or system* or program* or algorithm* or process* or method* or intervention* or material*)) or (decision adj2 (board* or guide* or counseling)) or “adaptive conjoint analys*”).mp.	396,116
21	or/15–20	398,913
22	14 and 21	17,169
23	exp meta analysis/	437,121
24	exp Meta-Analysis as Topic/	90,799
25	exp “systematic review”/	909,178
26	exp controlled study/	12,498,494
27	exp Randomized Controlled Trial/	1,808,422
28	exp triple blind procedure/	1349
29	exp Double-Blind Method/	678,327
30	exp Single-Blind Method/	150,076
31	exp latin square design/	568
32	((meta adj analys*) or (systematic* adj3 review*) or (control* adj3 study) or (control* adj3 trial) or (randomized adj3 study) or (randomized adj3 trial) or (randomised adj3 study) or (randomised adj3 trial) or “pragmatic clinical trial” or (doubl* adj blind*) or (doubl* adj mask*) or (singl* adj blind*) or (singl* adj mask*) or (tripl* adj blind*) or (tripl* adj mask*) or (trebl* adj blind*) or (trebl* adj mask*) or “latin square” or placebo* or nocebo* or random*).mp,pt.	19,380,524
33	or/23–32	19,381,772
34	22 and 33	5263
35	from 22 keep 11,508–17,169	5662
36	limit 35 to (controlled clinical trial or meta analysis or randomized controlled trial or pragmatic clinical trial or systematic reviews) [Limit not valid in APA PsycInfo,CCTR,CDSR,Embase; records were retained]	424
37	34 or 36	5263
38	limit 37 to (editorial or erratum or letter or note or addresses or autobiography or bibliography or biography or blogs or comment or dictionary or directory or interactive tutorial or interview or lectures or legal cases or legislation or news or newspaper article or overall or patient education handout or periodical index or portraits or published erratum or video-audio media or webcasts) [Limit not valid in APA PsycInfo,CCTR,CDSR,Embase,Ovid MEDLINE(R); records were retained]	189
39	from 38 keep 4–163	160
40	(37 not 38) or 39	5234
41	limit 40 to yr=“1860 - 2024”	4578
42	remove duplicates from 41	3561

Scopus

• 1 TITLE-ABS-KEY(“critical illness*” OR “critically ill” OR “end of life” OR “goals of care” OR hospice* OR “life-limiting” OR “serious illness*” OR “seriously ill” OR terminal OR “terminally ill”)

• 2 TITLE-ABS-KEY((symptomatic W/1 (treatment* or therap* or care or surg* or manag* or intervention* or radiotherap*)) or palliation or palliative)

• 3 TITLE-ABS-KEY(((advance or advanced) W/3 directive*) OR ((medical or healthcare or “health care” or durable) W/2 (power W/2 attorney)) OR “advance care” OR “advanced care” OR “living will” OR “living wills” OR “psychiatric will” OR “psychiatric wills” OR “ulysses contract*”)

• 4 1 or 2 or 3

• 5 TITLE-ABS-KEY(((decision* or decid*) W/2 (support* or aid* or tool* or instrument* or technolog* or technique* or system* or program* or algorithm* or process* or method* or intervention* or material*)) OR (decision W/2 (board* or guide* or counseling)) OR “adaptive conjoint analys*”)

• 6 TITLE-ABS-KEY((meta W/1 analys*) OR (systematic* W/3 review*) OR (control* W/3 study) OR (control* W/3 trial) OR (randomized W/3 study) OR (randomized W/3 trial) OR (randomised W/3 study) OR (randomised W/3 trial) OR “pragmatic clinical trial” OR (doubl* W/1 blind*) OR (doubl* W/1 mask*) OR (singl* W/1 blind*) OR (singl* W/1 mask*) OR (tripl* W/1 blind*) OR (tripl* W/1 mask*) OR (trebl* W/1 blind*) OR (trebl* W/1 mask*) OR “latin square” OR placebo* OR nocebo* OR random*)

• 7 4 and 5 and 6

• 8 DOCTYPE(le) OR DOCTYPE(ed) OR DOCTYPE(bk) OR DOCTYPE(er) OR DOCTYPE(no) OR DOCTYPE(sh)

• 9 7 and not 8

• 10 PMID(0*) OR PMID(1*) OR PMID(2*) OR PMID(3*) OR PMID(4*) OR PMID(5*) OR PMID(6*) OR PMID(7*) OR PMID(8*) OR PMID(9*)

• 11 9 and not 10

Appendix 3

Risk of bias/quality of all included studies.

Author, year	Random sequence generation	Allocation concealment	Blinding of participants and personnel		Blinding of outcome assessment		Incomplete outcome data	Selective reporting	Other sources of bias
	Risk	Risk	Risk	Yes/No	Risk	Yes/No	Risk	Risk	Risk
Allen 2018	Low	Low	High	Yes	Low	No	Low	Low	Moderate
Alsagheir 2020	Low	Unclear	Low	No	High	No	High	Low	Unclear
El-Jawahri 2010	Low	Low	Low	No	Unclear	No	Low	Low	Moderate
El-Jawahri 2015	Low	Low	Low	No	Low	No	Low	Low	Low
El-Jawahri 2016	Low	Low	High	Yes	Unclear	No	Low	Low	Moderate
Eneayna 2020	Low	Low	High	Yes	Unclear	No	Low	Low	Unclear
Hanson 2011	Low	Unclear	Low	No	Low	Yes	Low	Low	Unclear
Leighl 2011	Low	Low	Low	No	Unclear	No	Low	Low	Low
Matlock 2014	Low	Unclear	Low	No	High	Yes	High	Low	High
Meropol 2013	Low	Low	Low	Yes	Unclear	No	Unclear	Low	Moderate
Mcilvennan 2018	Low	Low	Low	No	Low	No	Low	Low	High
Oostendorp2017	Low	Low	Moderate	Partially	High	No	Low	Low	Moderate
Shirai 2011	Low	Unclear	High	No	High	No	Low	Low	Moderate
Stegmann 2020	Low	Low	Low	Yes	Low	No	Moderate	Low	Moderate
Volandes 2013	Low	Low	Low	No	High	No	Low	Low	Moderate
Yun 2011	Low	Unclear	Low	No	Low	Yes	High	Low	Unclear
Overall risk	Low	Moderate	High		Moderate- High		Low	Low	Moderate

Appendix 4

Figure 3a.

Forest plot for difference in quality of life between decision aid and control groups.

Quality of life assessed using a range of validated instruments; pooled as standardized mean differences.

Figure 3b.

Forest plot for difference in depression between decision aid and control groups.

Depression assessed using Hospital Anxiety and Depression Scale or Patient Health Questionnaire-2; pooled as standardized mean differences.

Figure 3c.

Forest plot for difference in values-related outcomes between decision aid and control groups.

Values clarity assessed using Decisional Conflict Scale subscales or study-specific values measures; pooled as standardized mean differences.

Figure 3d.

Forest plot for difference in anxiety between decision aid and control groups.

Anxiety assessed using Hospital Anxiety and Depression Scale, Generalized Anxiety Disorder–7 scale, Spielberger State–Trait Anxiety Inventory; pooled as standardized mean differences.

Data availability statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request.