Assessment of shared decision-making tool for eliciting informed goals of care in the hospitalized elderly (ASKmeGOC): protocol for a randomized clinical trial Authors: Chris Martin, Kelly Cruise, Monica Aggarwal, Doug Austgarden, Giulio DiDiodato Published: 2025 ABSTRACT AbstractBackgroundGoals of care discussions (GOCDs) occur between patients, substitute decision-makers, and healthcare providers to ensure shared decision-making about treatment preferences that are aligned with both patients’ values and goals and healthcare provider recommendations. GOCDs promote patient autonomy and high-value care. This study will examine the real-world effectiveness of using a novel ASKmeGOC e-tool (ASKmeGOC e-tool) to support nursing-facilitated GOCDs with patients and their substitute decision-makers. We hypothesize that these GOCDs will be acceptable to patients and attending physicians, promote more informed decisions about life-sustaining treatment preferences, and result in fewer undocumented CODE STATUS preferences that will result in both reduced rates of intensive care utilization and direct patient hospitalization costs compared to usual GOCDs conducted by physicians.MethodsThe ASKmeGOC trial is designed as a prospective, single-center, stratified, parallel group, allocation concealed, assessor masked, randomized, pragmatic, mixed-method, and comparative effectiveness trial. This study will enrol all consecutive hospitalized patients ≥ 80 years old. Patients will be randomized 1:1 to either the ASKmeGOC intervention or usual GOCD control group. The primary outcomes include total days of intensive care, ventilator and dialysis utilization during both the index hospitalization, and at 12 months post-admission. The hypothesized sample size is 950 participants per group to demonstrate a reduction in utilization of ≥ 20%. Count regression models will be used to analyze the primary outcomes using an intention-to-treat approach.DiscussionEnsuring patient autonomy in medical decision-making is an essential human right, and GOCDs are critical to ensure patient-centered medical care. This study will evaluate the real-world effectiveness of a novel tool to facilitate GOCDs by non-physician healthcare providers. If shown to be effective, the plan is to spread and scale the tool to other acute care hospitals, primary care practices, and assisted living facilities to evaluate the implementation effectiveness of its use in different healthcare environments by different healthcare providers with the goal of ensuring equitable access to standardized GOCDs to all eligible patients.Trial registrationClinicalTrials.gov NCT06002113. Registered on July 27, 2023.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13063-025-09323-6.Keywords: Facilitated goals of care discussions, Hospitalized patients, Life-sustaining treatment utilization, Decision aid, Decisional conflict, Mixed-method study Background Goals of care discussions (GOCDs) refers to the process of shared-decision making that occurs between a healthcare provider (HCP) and a patient and, ideally, their substitute decision-maker [1, 2]. The focus of GOCDs is to elicit care preferences that are informed by and aligned with a patient’s values and goals, while also taking HCP recommendations into consideration [3–6]. GOCDs make known the care preferences of patients and increase the likelihood that patients receive goal-concordant care [7] by promoting and protecting patient autonomy and minimizing unwanted or low-value care [8]. Many hospitalized patients with serious or deteriorating illnesses go on to receive aggressive medical and surgical interventions. Unlike terminal illnesses, it is difficult to predict which patients will benefit from these treatments [9, 10], making decisions about whether to accept (patient), provide (HCP), or withdraw (patient and HCP) life-sustaining therapies (LSTs) in these situations very complex and uncertain leading to tremendous variation in LST utilization [11, 12]. In addition, up to 75% of seriously ill patients will lack decision-making capacity at the time of their serious illness [13], with less than 25% having communicated their treatment preferences with either their substitute decision-maker or HCP. Failure to engage hospitalized patients with or at high risk of developing a serious illness may, therefore, lead to unwanted or low-value care and contribute to unnecessary pain and suffering in patients. This may also cause a negative emotional burden for substitute decision-makers who may doubt whether they had made the right decisions about LSTs [14, 15], thus emphasizing the need for high-quality upstream GOCDs in high-risk populations, prior to deterioration or development of a serious illness. Some have even suggested that this failure to engage high-risk hospitalized patients in GOCDs is tantamount to a medical error of omission that adversely affects patient safety [16]. Even in those patients who have been approached by HCPs about their preferences, the language used in these conversations is critical and can lead to significant misunderstandings between patients and HCPs that may promote goal-discordant care [17–21]. Given the multidimensional and complex nature of GOCDs, it is not surprising that there exist many barriers to effective communication between HCPs and patients [3, 21–34] resulting in many missed opportunities for elicitation of goals of care in seriously ill hospitalized patients [35]. The most commonly reported barriers are lack of knowledge about outcomes and prognoses associated with acute illnesses, lack of knowledge about risks, benefits and outcomes associated with care options and preferences, and communication training and time to commit to a GOCD. In response, many recent studies and recommendations to promote GOCDs have been published [6, 29, 33, 34, 36–41]. While providing useful guides to promote conversations that contain important components of a comprehensive GOCD, they are limited by the general nature of their content, inability to personalize information about care options, outcomes, and prognoses, and variability in their adoption and implementation in clinical practice. Few of the tools are disease agnostic, therefore limiting their scalability and applicability to a general hospitalized population at high risk of dying from any cause and not necessarily from the terminal progression of a specific chronic illness. For all the guides and tools, there was a lack of consistent patient outcomes evidence to recommend the clinical use of any one intervention compared to another. There was a lack of evidence to support how best to implement these interventions in a systematic way across the healthcare system to ensure a standardized approach to GOCDs, and this resulted in very few tools being adopted and implemented as a part of the routine care of seriously ill hospitalized patients. The difficulty of translating patient’s values and goals into care options and recommendations was frequently encountered with few tools equipped to facilitate this component of GOCDs. Many of the tools were not publicly available, and many neglected to use evidence-based techniques for presenting outcome and prognostic information. Because most tools were paper based, the content could not be easily personalized for individuals’ specific set of clinical circumstances, and they could not be easily updated. While all of the computer-based tools tested in randomized clinical trials demonstrated efficacy to improve at least one patient-centered outcome such as decisional conflict, satisfaction with decisions, or health care outcomes, all were directed to specific chronic disease conditions, and none was generalizable to seriously ill hospitalized patients. In response, we developed a digital tool (ASKmeGOC e-tool) to promote GOCDs in a general hospital population and tested feasibility of implementation, intervention fidelity, and patient and healthcare provider acceptability [42]. Our pilot study demonstrated acceptable implementation, intervention adherence, and acceptability among stakeholders and suggested there may be an important benefit in ensuring that treatment preferences for LSTs are goal-concordant and documented across the healthcare system. In that pilot, the ASKmeGOC e-tool was used by a physician assistant to deliver a facilitated GOCD (f-GOCD). Our plan is to conduct a randomized proof-of-concept study to verify the findings from the feasibility study. We will randomize patients to f-GOCDs delivered by nurses using the ASKmeGOC e-tool compared to usual care to determine if there is evidence of effectiveness in reducing utilization of LSTs. Methods Aims/objectives To determine the impact of f-GOCDs on the number of intensive care unit (ICU), ventilator, and dialysis days during the index hospitalization To determine the impact of f-GOCDs on the number of intensive care unit (ICU), ventilator, and dialysis days during the index hospitalization To determine the impact of f-GOCDs on the number of ICU, ventilator, and dialysis days after the index hospitalization until 12 months after admission (or until death) To determine the impact of f-GOCDs on the number of ICU, ventilator, and dialysis days after the index hospitalization until 12 months after admission (or until death) To determine the impact of f-GOCDs on the final treatment preferences for life-sustaining treatments documented in CODE STATUS, where CODE STATUS represents the emergency care treatment plans documented in an electronic or paper patient health record specifying goals of care directives To determine the impact of f-GOCDs on the final treatment preferences for life-sustaining treatments documented in CODE STATUS, where CODE STATUS represents the emergency care treatment plans documented in an electronic or paper patient health record specifying goals of care directives To determine the impact of f-GOCDs on other outcomes including decisional conflict and quality of communication, patient satisfaction with the encounter, and place of death To determine the impact of f-GOCDs on other outcomes including decisional conflict and quality of communication, patient satisfaction with the encounter, and place of death To determine the difference in direct patient hospital costs To determine the difference in direct patient hospital costs To determine the barriers and facilitators to the implementation of GOCDs To determine the barriers and facilitators to the implementation of GOCDs Design The ASKMEGOC trial is designed as a prospective, single-center, stratified, parallel group, allocation concealed, assessor masked, randomized, pragmatic, mixed-method, and comparative effectiveness trial in hospitalized elderly patients 80 years and older. Sample size In our pilot study, there was an absolute difference of 38.2% (95% CI 22.1% to 54.3%) fewer patients in the f-GOCD group whose final CODE STATUS included LSTs compared to the control group [42]. We hypothesized that a ≥ 20% reduction in LSTs days is clinically significant. A Poisson regression of the composite outcome of LST utilization for the index hospitalization on a binomial distributed independent variable (f-GOCD vs usual care) using a total sample of 800 observations would achieve power ≥ 80% at the Bonferroni-corrected = 2.5% significance level for a two-sided test to detect a reduction in the composite outcome of LST utilization of ≥ 20% due to the f-GOCD intervention if the baseline composite LST utilization rate was 0.017 LST days per hospital day, and the mean hospital length of stay was 12 days. These assumptions were derived from the Canadian Institute for Health Information (estimated average length of stay in elderly patients ≥ 80 years was 12.27 days (95% CI 11.9, 12.6)) [43], and the average composite outcome of LST utilization rate of 0.017 days per hospital day was derived from our own hospital’s statistics for the 5-year period from 2016 to 2020 (mean = 0.0169 (SD 0.0039)). Other modifications to the estimated sample size include the following reasons: (i)We demonstrated in our pilot study that 75% of eligible patients consented and were enrolled in the study, so requiring an increase in the sample size to 535 patients for each group to account for this no consent group. We demonstrated in our pilot study that 75% of eligible patients consented and were enrolled in the study, so requiring an increase in the sample size to 535 patients for each group to account for this no consent group. (ii)Only a minority of patients (11%) that were eligible in the pilot study were exposed to the f-GOCD due to time constraints. In that pilot study, a single physician assistant was responsible for all f-GOCD, including their other usual clinical duties. In the current study, intensive care unit study nurses will be conducting the f-GOCD, and during the study periods, they will have no other competing duties. Despite this, we estimate that up to 25% of all eligible patients may still be missed [44] given the study period only covers from Monday-Friday during the hours 9:00 to 17:00, with all holidays excluded. This would require a further increase in our sample from 535 to 713 patients per group to account for this missed proportion. Only a minority of patients (11%) that were eligible in the pilot study were exposed to the f-GOCD due to time constraints. In that pilot study, a single physician assistant was responsible for all f-GOCD, including their other usual clinical duties. In the current study, intensive care unit study nurses will be conducting the f-GOCD, and during the study periods, they will have no other competing duties. Despite this, we estimate that up to 25% of all eligible patients may still be missed [44] given the study period only covers from Monday-Friday during the hours 9:00 to 17:00, with all holidays excluded. This would require a further increase in our sample from 535 to 713 patients per group to account for this missed proportion. (iii)In our pilot study, we demonstrated that only 2.07% (95% CI 1.16 to 3.38) of nonexposed patients (usual care group) did not complete a CODE STATUS prior to discharge or death. A recent review of CODE STATUS completion rates among patients matching the study eligibility criteria demonstrated approximately 75% completion rates, necessitating a further increase in sample size from 713 to 950 patients in both groups. In our pilot study, we demonstrated that only 2.07% (95% CI 1.16 to 3.38) of nonexposed patients (usual care group) did not complete a CODE STATUS prior to discharge or death. A recent review of CODE STATUS completion rates among patients matching the study eligibility criteria demonstrated approximately 75% completion rates, necessitating a further increase in sample size from 713 to 950 patients in both groups. Taking all these additional factors into account, and given the allocation ratio will be 1:1, we would require randomizing approximately 1900 patients overall. This sample size would have statistical power ≥ 90% at the = 2.5%) to detect the same effect size for the other primary outcome (LST and hospital utilization in 12 months following admission from index hospitalization), even accounting for a composite outcome rate 10-fold lower than the previously estimated LST rate of 0.017 days per hospital day and the mean follow-up time being only 160 days. Setting This study will be conducted at the Royal Victoria Regional Health Centre in Barrie, ON, Canada. The Royal Victoria Regional Health Care Center is a 388-bed acute care, large community hospital serving a municipality of 150,000 residents located 100 km north of Toronto, ON, in Central Ontario. Inclusion criteria Hospitalized patients ≥ 80 years old with an acute medical or surgical condition admitted to any hospital ward Hospitalized patients ≥ 80 years old with an acute medical or surgical condition admitted to any hospital ward Duration of admission ≥ 24 h Duration of admission ≥ 24 h No CODE STATUS documented or documented CODE STATUS includes preferences for LSTs No CODE STATUS documented or documented CODE STATUS includes preferences for LSTs English speaking or translator present English speaking or translator present Competent patient or substitute decision-maker provides verbal consent Competent patient or substitute decision-maker provides verbal consent Exclusion criteria Patient, or substitute decision-maker, declines Patient, or substitute decision-maker, declines Previously or currently documented CODE STATUS preferences do not include any LSTs Previously or currently documented CODE STATUS preferences do not include any LSTs New diagnosis of life-limiting illness on this hospital admission, for example, new diagnosis of metastatic cancer New diagnosis of life-limiting illness on this hospital admission, for example, new diagnosis of metastatic cancer Clinically unstable, admitted to an intensive care unit, or currently receiving acute life support treatment (mechanical ventilation, acute dialysis, or inotropic/vasopressor support) Clinically unstable, admitted to an intensive care unit, or currently receiving acute life support treatment (mechanical ventilation, acute dialysis, or inotropic/vasopressor support) Readmission after index hospitalization Readmission after index hospitalization Preexisting need for chronic mechanical ventilation (invasive mechanical ventilation via tracheostomy > 90 days) Preexisting need for chronic mechanical ventilation (invasive mechanical ventilation via tracheostomy > 90 days) Interventions All eligible patients will be allocated to one of two parallel groups (Fig. 1). Permuted block design will be used to randomly allocate participants to either f-GOCD intervention or usual care in a 1:1 ratio within each stratum. Allocation will be stratified by the following: (i)Whether a resuscitation level designation form had been completed in the preceding 12 months (yes versus no) Whether a resuscitation level designation form had been completed in the preceding 12 months (yes versus no) (ii)Medical versus surgical admission Medical versus surgical admission (iii)Sex (male versus female): The blocks will be of variable sizes. There will be no a priori fixed ratio for allocation in the different strata. The randomization table will be created by an independent statistician using the ralloc command in STATA/MP 16.1 for Mac (https://www.stata.com/). The randomization table will be uploaded into a secure, web-based software platform (REDCap®) hosted at the Royal Victoria Regional Health Centre. The allocation sequence will be concealed from researchers responsible for screening eligibility and enrolment. To enrol patients once they were deemed eligible, these researchers will access the randomization module in REDCap® to assign patients to either the f-GOCD intervention or usual care. Sex (male versus female): The blocks will be of variable sizes. There will be no a priori fixed ratio for allocation in the different strata. The randomization table will be created by an independent statistician using the ralloc command in STATA/MP 16.1 for Mac (https://www.stata.com/). The randomization table will be uploaded into a secure, web-based software platform (REDCap®) hosted at the Royal Victoria Regional Health Centre. The allocation sequence will be concealed from researchers responsible for screening eligibility and enrolment. To enrol patients once they were deemed eligible, these researchers will access the randomization module in REDCap® to assign patients to either the f-GOCD intervention or usual care. Consort flow diagram depicting study phases f-GOCD intervention The f-GOCD intervention has been previously described in our pilot study [42]. In brief, a locally developed web-based, ASKmeGOC e-tool will provide a structured approach to conducting f-GOCDs with hospitalized patients at high risk of serious illness. Different from the pilot study, ICU nurses use the ASKmeGOC e-tool to conduct the f-GOCD. Preemptive consent to approach eligible patients for a f-GOCD was obtained from the department chiefs for all attending physicians. Each morning at 9:00 am from Monday to Friday, an ICU study nurse will run a study report that identifies all consecutive patients ≥ 80 years old admitted to a hospital ward in the previous 24 h (or from Friday 09:00 to Monday 09:00 on Monday mornings since there is no coverage on either Saturday or Sunday). The ICU study nurse will then verify eligibility by using the electronic health record and/or contacting the primary care team as required. All eligible patients will undergo Zelen randomization [45] by the research coordinator. The ICU study nurse will approach the patient and/or the substitute decision-maker to obtain verbal consent to conduct a f-GOCD using the ASKmeGOC e-tool. If consent is provided, a scheduled time with both the patient and/or the substitute decision-maker will attempt to be organized for the f-GOCD. Ideally, all f-GOCDs will take place with both of these agents to ensure that the substitute decision-maker is fully aware of the patient’s treatment preferences. The goal is to ensure that capable at-risk patients are making their own informed decisions well in advance of a life-threatening illness. This will not only ensure goal-concordant care in the event of deterioration but will reduce the decision-making burden of substitute decision-makers, as patient wishes have already been expressed. The ASKmeGOC e-tool has four major components that include the following: (i) Prehospital assessment of quality of life and frailty, (ii) prognostic score for expected hospital survival rate given admission diagnosis, (iii) prognostic score for expected hospital survival rate after inhospital cardiopulmonary arrest, and (iv) elicitation of patients’ values and goals and alignment with HCP recommendations. Once completed, a formal review of the patient’s resuscitation treatment preferences will take place, and, if desired, the ICU study nurse will help the patient/substitute decision-maker complete or revise their CODE STATUS. The ICU study nurse will then review the f-GOCD and the CODE STATUS with the study intensivist. The study intensivist will document their opinion (agreement or disagreement) about the patient’s CODE STATUS. Any disagreement will be resolved by the study intensivist discussing the GOCD contents with the attending physician/patient/substitute decision-maker. Following this review, the study intensivist will contact the attending physician and inform them of any changes to the CODE STATUS. The attending physician will be asked to provide their opinion (agreement or disagreement) about the patient’s CODE STATUS. In our pilot study [42], no attending physician disagreed with the patient’s/substitute decision-maker’s preferences elicited through the intervention. The ICU study nurse will document the outcome of the f-GOCD using a standardized reporting template along with updating the patient’s CODE STATUS in the patient electronic medical record (EMR). Both will be available to any healthcare provider with access to the EMR, including the patient’s family physician in the community who will receive a copy of the EMR report of the f-GOCD summary. Usual care In the usual care arm (control or declined f-GOCD intervention), attending physicians may choose to conduct GOCDs as deemed necessary using whatever approach they deem appropriate. Attending physicians may choose to complete the CODE STATUS with the patient and/or their substitute decision-maker as deemed necessary using whatever approach they deem appropriate. Attending physicians will not be able to access the ASKmeGOC e-tool to facilitate their GOCDs during the study period nor will they be able to request a f-GOCD from the study nurses to prevent contamination. Patients who do not complete a CODE STATUS are, by default, to receive full resuscitative measures at the Royal Victoria Regional Health Centre. Outcomes Primary A composite outcome of the sum (= total) of ICU, ventilator, and dialysis days during the index hospitalization (or until death in the index hospitalization) will be calculated. The total number of days a patient is alive during the index hospitalization will be recorded. These counts will represent the risk period (exposure). ICU days represent the number of days spent in an ICU (in whole or in part where the beginning of each calendar day occurs at 00:00 and the end of each calendar day occurs at 23:59). For example, if a patient was admitted to an ICU at 23:00 on February 1, 2021, and transferred to the ward at 09:00 on February 4, 2021, the number of ICU days recorded would be 4 days. Ventilator and dialysis days will be similarly counted. Sensitivity analysis using LSTs-free days will also be determined and compared between groups to account for any bias due to informative censoring from death. A composite outcome of the sum (= total) of ICU, ventilator, and dialysis days during the index hospitalization (or until death in the index hospitalization) will be calculated. The total number of days a patient is alive during the index hospitalization will be recorded. These counts will represent the risk period (exposure). ICU days represent the number of days spent in an ICU (in whole or in part where the beginning of each calendar day occurs at 00:00 and the end of each calendar day occurs at 23:59). For example, if a patient was admitted to an ICU at 23:00 on February 1, 2021, and transferred to the ward at 09:00 on February 4, 2021, the number of ICU days recorded would be 4 days. Ventilator and dialysis days will be similarly counted. Sensitivity analysis using LSTs-free days will also be determined and compared between groups to account for any bias due to informative censoring from death. A composite outcome of the sum (= total) of ICU, ventilator, and dialysis days during the 12-month period following admission from the index hospitalization (or until death). The total number of days a patient is alive during this 12-month period will be recorded and represent the risk period (exposure). ICU days represent the number of days spent in an ICU (in whole or in part where the beginning of each calendar day occurs at 00:00 and the end of each calendar day occurs at 23:59). Ventilator and dialysis days will be similarly counted. Sensitivity analysis using LSTs-free days will also be determined and compared between groups to account for any bias due to informative censoring from death. A composite outcome of the sum (= total) of ICU, ventilator, and dialysis days during the 12-month period following admission from the index hospitalization (or until death). The total number of days a patient is alive during this 12-month period will be recorded and represent the risk period (exposure). ICU days represent the number of days spent in an ICU (in whole or in part where the beginning of each calendar day occurs at 00:00 and the end of each calendar day occurs at 23:59). Ventilator and dialysis days will be similarly counted. Sensitivity analysis using LSTs-free days will also be determined and compared between groups to account for any bias due to informative censoring from death. The primary outcomes will be reported independently, and clinical significance is not dependent on both demonstrating an outcome in the same direction. For example, if LST utilization is decreased in the ASKmeGOC group during the index hospitalization, but is not different between groups at the 12-month period, this may simply reflect that the effect of the ASKmeGOC intervention dissipates over time, which has been an observed finding in previous studies. Secondary Compare the differences in the distributions for LSTs preferences documented in the CODE STATUS at discharge (or death) during the index hospitalization. In the current study, the treatment categories in the CODE STATUS include the following: (i) Comfort care (equivalent to palliation — no LSTs); (ii) supportive/medical care (ongoing medical and/or surgical treatment but no LSTs); (iii) minimally invasive LSTs, such as inotropes/vasopressors, but no cardiopulmonary resuscitation (DNACPR); (iv) invasive LSTs, such as invasive mechanical ventilation, but DNACPR; and (v) any LST including CPR. An undocumented CODE STATUS defaults to (v) any LST including CPR. Compare the differences in the distributions for LSTs preferences documented in the CODE STATUS at discharge (or death) during the index hospitalization. In the current study, the treatment categories in the CODE STATUS include the following: (i) Comfort care (equivalent to palliation — no LSTs); (ii) supportive/medical care (ongoing medical and/or surgical treatment but no LSTs); (iii) minimally invasive LSTs, such as inotropes/vasopressors, but no cardiopulmonary resuscitation (DNACPR); (iv) invasive LSTs, such as invasive mechanical ventilation, but DNACPR; and (v) any LST including CPR. An undocumented CODE STATUS defaults to (v) any LST including CPR. Compare the proportions of patients who change their CODE STATUS during the index hospitalization and the distribution of these changes (escalation versus de-escalation). Compare the proportions of patients who change their CODE STATUS during the index hospitalization and the distribution of these changes (escalation versus de-escalation). Compare the difference in time (hours) between the date of admission and the completion of the first CODE STATUS. Compare the difference in time (hours) between the date of admission and the completion of the first CODE STATUS. Time required to complete f-GOCD Time required to complete f-GOCD Quality of goals and care discussion: For each group, the quality of goals of care discussions will be evaluated using the 13-item Quality of Communication questionnaire [46]. The questionnaire is a validated instrument used to assess patient-provider communication in seriously ill hospitalized patients. The questionnaire can be completed by either patients or their surrogate decision-makers involved in a GOCD. The questionnaire consists of a 6-item general communication scale (QOC-GEN) and a 7-item end-of-life communication scale (QOC-EOL). Total scores for each subscale will be calculated and compared. Quality of goals and care discussion: For each group, the quality of goals of care discussions will be evaluated using the 13-item Quality of Communication questionnaire [46]. The questionnaire is a validated instrument used to assess patient-provider communication in seriously ill hospitalized patients. The questionnaire can be completed by either patients or their surrogate decision-makers involved in a GOCD. The questionnaire consists of a 6-item general communication scale (QOC-GEN) and a 7-item end-of-life communication scale (QOC-EOL). Total scores for each subscale will be calculated and compared. Patient decision satisfaction defined as the absence of decisional conflict. Decisional conflict will be measured using the 4-item SURE questionnaire [47]. The questionnaire is a validated instrument used to assess clinically significant decisional conflict among patients in clinical practice. The total score for the questionnaire is 4, and any score ≤ 3 has been shown to be associated with decisional conflict. Patient decision satisfaction defined as the absence of decisional conflict. Decisional conflict will be measured using the 4-item SURE questionnaire [47]. The questionnaire is a validated instrument used to assess clinically significant decisional conflict among patients in clinical practice. The total score for the questionnaire is 4, and any score ≤ 3 has been shown to be associated with decisional conflict. Compare the proportion of patients that did not have a documented CODE STATUS during their index hospitalization at the time of discharge or death. Compare the proportion of patients that did not have a documented CODE STATUS during their index hospitalization at the time of discharge or death. Measure the proportion of eligible patients in the f-GOCD group that did not receive the intervention. Measure the proportion of eligible patients in the f-GOCD group that did not receive the intervention. Compare the total direct hospital costs of the index hospitalization. Compare the total direct hospital costs of the index hospitalization. Evaluation of the ASKmeGOC e-tool using a 7-item evaluation instrument previously used to evaluate a novel advanced care planning tool [48]. The implementation of the ASKmeGOC e-tool can present challenges to existing work processes and impact on care delivery, which can be overlooked in the intervention design. Thus, it is important to understand the factors that will effectively support the implementation of the ASKmeGOC e-tool. We will use the Consolidated Framework for Implementation Research (CFIR) to study the barriers and facilitators to using the ASKmeGOC e-tool. CFIR is a well-recognized conceptual framework that is used to guide systematic evaluations of multilevel factors that shape the implementation and effectiveness of interventions [49]. This study will qualitatively explore experiences of health care practitioners, patients, and their substitute decision-makers to understand their engagement with the intervention in order to understand how to better implement the ASKmeGOC e-tool. Evaluation of the ASKmeGOC e-tool using a 7-item evaluation instrument previously used to evaluate a novel advanced care planning tool [48]. The implementation of the ASKmeGOC e-tool can present challenges to existing work processes and impact on care delivery, which can be overlooked in the intervention design. Thus, it is important to understand the factors that will effectively support the implementation of the ASKmeGOC e-tool. We will use the Consolidated Framework for Implementation Research (CFIR) to study the barriers and facilitators to using the ASKmeGOC e-tool. CFIR is a well-recognized conceptual framework that is used to guide systematic evaluations of multilevel factors that shape the implementation and effectiveness of interventions [49]. This study will qualitatively explore experiences of health care practitioners, patients, and their substitute decision-makers to understand their engagement with the intervention in order to understand how to better implement the ASKmeGOC e-tool. For each patient who dies during the study period, the date and place of death will be collected. Place of death will be defined as outside hospital, hospital-not ICU, and hospital-ICU. For each patient who dies during the study period, the date and place of death will be collected. Place of death will be defined as outside hospital, hospital-not ICU, and hospital-ICU. Timelines(Fig. 2) Spirit template for scheduled study enrolment, interventions, and assessments. *Abbreviations (in order of appearance): f-GOCD, facilitated goals of care discussion; HSMR, hospital standardized mortality rate; VAS, visual analog scale; LST, life-sustaining treatment; CRA, cardiorespiratory arrest; EMR, electronic medical record Data management Only authorized trial personnel will have password-protected access to the study data. All study data will be stored according to the Personal Health Information Protection Act (PHIPA) [50] in REDCap [51]. All trial personnel who work with this data will have their Tri-Council Policy Statement 2 certification in the ethical conduct for research involving humans. A randomly generated unique study identifier will be assigned to each study participant. This study identifier will be used exclusively in all analytical files. This study identifier will be linked to a unique patient health identifier in a master list that will be used throughout the study duration. This master list will be maintained by the principal investigator in a password-protected Excel file on a dedicated laptop computer in the principal investigator’s private office to ensure file security. This file will be made available to other trial personnel as needed for the sole purpose of conducting the study. All study data will be maintained for 15 years and subsequently permanently destroyed. The minimum amount of personal health information will be collected to limit the risk of data breaches. All personal identifying information will be removed from REDCap once enrolment and follow-up are complete. Direct and indirect personal health identifiers will not be included in the analytic database. Participant encounters by trial personnel will take place in private areas only to avoid risk of f-GOCDs being overheard by others not directly involved in the research. Follow-up phone calls (if applicable) will be conducted by trial personnel from private offices to avoid being overheard by anyone who is not directly involved with the research. Voicemail messages will not be left for participants who are not available by phone. Safety The principal investigators have extensive experience with the f-GOCD intervention, having created the software, tested it on volunteers, trained healthcare providers (study ICU nurses and physician assistants) on how to incorporate the tool in a structured approach to f-GOCDs, and used it to pilot test its effectiveness in a similar population included in this trial [42]. During the pilot, there were no reported adverse effects among the exposed patients, substitute decision-makers, or their attending physicians. Additional protective measures include trial personnel certified in human subjects’ research ethics and regulations, voluntary participation of patients, and substitute decision-makers through an informed consent process, PHIPA-compliant data collection, storage and use, study monitor, and data monitoring committee providing independent audit and trial oversight. Statistical analysis plan LST utilization We will use zero-inflated count multivariate regression model to assess the primary composite outcome of LST utilization during the index hospitalization and the 12-month period from the time of admission of the index hospitalization. The rationale for this modelling is based on review of the historical ICU utilization data previously mentioned. Sensitivity analysis assessing the potential impact of overdispersion on the model outcomes will be included. Only the final CODE STATUS categories will be forced into the final model, as all other variables are expected to be balanced given the RCT design. The risk period for the index hospitalization will be defined as the hospital length of stay, and the risk period for the 12-month period following admission for the index hospitalization will be defined as the number of days alive (up to a maximum of 365.25 days from the time of admission of the index hospitalization). The main analyses will be conducted on an intention-to-treat basis by analyzing participants according to the treatment to which they were assigned and regardless of compliance. Baseline data and secondary outcomes For secondary outcomes that report proportions, 2 analyses will be used to compare groups. For secondary outcomes that report scores from questionnaires, the ANOVA or Wilcoxon rank-sum test will be used to compare groups depending on the underlying distribution of scores. We will compare the demographic and baseline values of all variables across the treatment arms using ANOVA and 2 tests for continuous and binary data, respectively, and include those variables in a sensitivity analysis in the final statistical model whose p-value < 0.001 for differences across groups. For descriptive analyses, means and 95% confidence intervals (CI) or medians and interquartile ranges will be used to summarize observations depending on the underlying distribution of data. All categorical data will be reported as frequencies and proportions. All secondary outcome differences are exploratory. During the qualitative study, data collection and analysis will take place concurrently. The transcripts will be analyzed using techniques of qualitative thematic analysis, inspired by a deductive directed approach, since the CFIR constructs is being used as an a priori analysis framework. Transcripts will be coded using NVivo 10 qualitative analysis software. Thematic analysis of data will take place using the constant comparative method. To ensure that there is inter-rater reliability, a member of the team will read a sample of transcripts to confirm the accuracy and reliability of the coding. If the comparison of coding is not within 95%, researchers will work together to build consensus on the coding structure. All interviews will be reviewed and re-coded based on this structure. Discussion between reviewers will be ongoing to ensure the qualitative analysis is reliable and accurate. Missing data No imputation procedures will be used to replace missing data. All patients, regardless of missing data, will be included in the final analyses. All outcome data will be collected for all randomized patients to the intervention group regardless if they provide consent for the intervention. The rationale for this is that all outcome data is being collected in patients regardless of their inclusion in the study given this is routinely collected data during their hospitalization. This practice is consistent with pragmatic studies. Patient and public involvement Patients and the public were involved in the design of the original version and subsequent modified versions of the ASKmeGOC e-tool [42]. Discussion In general, GOCD should include the following four component parts [3, 4, 6, 36, 52]: (i)A review of the current medical and surgical condition responsible for hospitalization along with the prognosis and care options to ensure patient understanding of their clinical condition A review of the current medical and surgical condition responsible for hospitalization along with the prognosis and care options to ensure patient understanding of their clinical condition (ii)An elicitation of the patient’s values and priorities regarding care that is anchored at one end by a preference to extend life at all costs and on the other end to focus only on comfort care to be used later to guide treatment preferences and decisions An elicitation of the patient’s values and priorities regarding care that is anchored at one end by a preference to extend life at all costs and on the other end to focus only on comfort care to be used later to guide treatment preferences and decisions (iii)Translation of these values and priorities into available treatment options and their trade-offs informed by HCP clinical expertise and evidence for benefit Translation of these values and priorities into available treatment options and their trade-offs informed by HCP clinical expertise and evidence for benefit (iv)Documentation of the agreed-upon treatment preferences that also include informed decisions about life-sustaining treatments such as intensive care support and cardiopulmonary resuscitation. Documentation of the agreed-upon treatment preferences that also include informed decisions about life-sustaining treatments such as intensive care support and cardiopulmonary resuscitation. There is a single study that attempted to validate the value of these components in promoting goal-concordant care [53]. Prior to their validation, Sinuff et al. had developed a set of quality indicators using a Delphi method to evaluate GOCDs that included these four components [54]. In their final list of 13 quality indicators for GOCDs, component (i) was ranked as the most important among the indicators with a final mean score of 6.75 (standard deviation (SD) 0.44) (maximum importance score of 7 on a Likert scale). Component (iii) was ranked second (mean score 6.63 (SD 0.65)), component (ii) was ranked sixth (mean score 6.29 (SD 0.81)), and component (iv) was ranked ninth (mean score 6.21 (SD 1.02)). In the documentation category, documentation of the GOCDs in the medical record was ranked second overall among the 32 quality indicators they developed to evaluate the entire spectrum of end-of-life communication and decision-making. In their validation study [53, 55], Heyland et al. enrolled 297 hospitalized patients and 209 family members at 12 Canadian teaching hospitals. The medical records and participants were reviewed to determine if any of the quality indicators had been part of a serious illness conversation. The investigators assigned the importance scores derived from the Delphi study to each quality indicator that had been affirmed to be part of a serious illness conversation. They derived an overall quality indicator score by summing the importance scores of all the quality indicators included in the serious illness discussion and then standardized this value on a scale between 0 and 100% by dividing each sum by the raw sum of highest possible values of these items. The quality indicator scores for GOCDs ranged between 5 and 31%, suggesting that the quality of GOCDs occurring in these hospitals with these patients and families was both highly variable and very poor. Even the documentation scores were very low, ranging from 42 to 78%. The correlation between the GOCD quality indicator scores and concordance between stated preferences and documented goals of care was moderate but not statistically significant (r = 0.53, p = 0.08). The correlation between the ACP quality indicator scores and concordance between stated preferences and documented goals of care was even weaker (r = 0.35, p = 0.3), reinforcing the importance of having real-time GOCDs with hospitalized patients [56]. Identifying hospital inpatients at high risk of clinical deterioration is an essential first step in targeting communication about goals of care. Many studies have identified high-risk patients for targeting GOCDs through the use of the “surprise” question (“Would I be surprised if this patient died in the next year?”) [57]. Originally designed to identify cancer patients who might benefit from a palliative care consult, its use has been extended to include non-cancer patients. This systematic review evaluated the performance characteristics of the surprise question in 16 studies involving over 11,621 patients. The investigators found moderate (n = 10) to high (n = 4) risk of overall bias in 14 studies. The discriminative ability of the surprise question for predicting death at 6 to 12 months was modest in cancer patients (area under summary receiver operating curve 0.83 (95% CI 0.79, 0.87)) and poor in non-cancer patients (area under summary receiver operating curve 0.77 (95% CI 0.73, 0.81)). The surprise question had no role in predicting death during an acute hospitalization. Other studies have included patients between the ages of 55 and 79 years of age with ≥ 1 end-stage organ failure [52, 58–60]. While the rationale to support inclusion of this heterogenous group of patients was never provided in these studies, their face validity appears to be consistent with clinical experience and the natural history of their chronic diseases’ illness trajectories. In our pilot study, we targeted all hospitalized inpatients 80 years and older [42]. From 2014 to 2018, elderly patients 80 years and older accounted for 17.9% (95% CI 17.66, 18.14) of all inpatient hospitalizations despite making up only 4.23% (95% CI 4.15, 4.31) of the Canadian population [43]. Compared to elderly inpatients 65 to 79 years old, the age-specific hospital admission rates for elderly inpatients 80 years and older were 2.12-fold greater (95% CI 2.09, 2.14). These patients accounted for 29.37% (95% CI 28.93, 29.81) of all inpatient hospital days, and their age-specific length of stay was 12.27 days (95% CI 11.93, 12.60) or 1.38-fold longer (95% CI 1.37, 1.39) than elderly inpatients 65 to 79 years old. At the Royal Victoria Regional Health Centre, elderly patients 80 years and older accounted for 49.24% (95% CI 48.11, 50.37) of all inpatient deaths from 2015 to 2019 (personal communication, decision support unit, Royal Victoria Regional Health Centre). Among hospitalized elderly patients 75 years and older, this group accounts for the largest proportion of ICU admissions (30.5% of all ICU admissions) [61]. Among hospitalized patients with an intensive care admission, this group also accounts for the largest proportion of deaths following ICU admission (49.1% of all post-ICU deaths). In a multicenter, cluster-randomized clinical trial examining survival outcomes for 3036 elderly (≥ 75 years old) patients with a serious illness admitted to hospitals that had been randomized to either a systematic strategy promoting ICU admission or usual care, there was no difference in survival at 6 months despite higher rates of ICU admission in the ICU strategy hospitals (relative risk = 1.68 (95% CI 1.54, 1.82)) [62]. The elderly patients included in this study had to have preserved functional status as measured by the Index of Independence in Activities of Daily Living score ≥ 4 and preserved nutritional status, thus excluding elderly patients with frailty which is a known risk factor for poor ICU outcomes [63, 64]. Regardless of the reasons for hospitalization, the relative risk for death post-hospitalization in the elderly 80 years and older exceeds that compared to population-matched community-dwelling adults and continues to increase with time post-discharge [65]. There is no consensus as to who should be responsible for initiating and navigating GOCDs in seriously ill hospitalized patients [66]. If GOCDs result in treatment decisions about providing or withholding life-sustaining treatments, then physicians would have to be involved to attest and sign life-sustaining treatment forms. Many studies have examined the use of trained, non-physician facilitators to conduct GOCD with patients [67–74]. In general, these studies have demonstrated improvement in goal-concordant hospital care. When asked about their willingness to engage in GOCDs, nurses were less willing to initiate GOCDs (score of 5 on a scale of 7, 7 = “extremely willing”) compared to both residents and physicians [22]. Nurses consistently rated the importance of barriers to GOCDs higher than both residents and physicians, with the most important barriers identified as discordance between patients’ and families’ and HCPs’ expectations for outcomes, expectations for care options, and goals of care. Nurses did not feel particularly supported to engage in GOCDs by their organizations (score of 4 on a scale of 7, 7 = “extremely supported”). When physicians were asked how acceptable it was for non-physician HCPs to engage in GOCDs, their ratings ranged from as low as 3 to 6 (on a scale of 7, 7 = “extremely acceptable”). Physicians found it to be most acceptable for nurse practitioners and nurses to initiate GOCDs and least acceptable for social workers and other non-physician HCPs. However, when it came to making a final decision about care options, physicians were much less accepting of any non-physician HCP being involved (range 2 to 5, 1 = “extremely unacceptable & 7 = “extremely acceptable”). This pattern of acceptability for different components of GOCDs was also mirrored by nurses’ responses with highest acceptability for initiating GOCDs, followed by decreasing acceptability to exchanging information, being a decision coach and making a final decision. However, nurses were equally accepting of nurse practitioners, nurses, social workers, and other non-physician HCPs engaging in any of the components of GOCDs. There is no consensus about the ideal time to approach hospitalized patients for GOCDs [75]. In general, patients’ preferences about LSTs are known to vary over time and context. This may explain why wishes for goals of care identified in advanced care plans (ACPs) are poorly predictive of receiving goal-concordant care at some later time in the future during an acute illness [76–78]. Many ACPs are completed far in advance of the time of death without evidence of regular review and updating, bringing into question how valid they may be in representing the patient’s current wishes and priorities for care [79]. As a result, regardless of the availability of ACPs, all hospitalized patients at high risk of serious illness should have a GOCD [80]. Ideally, these GOCDs should not occur in unstable patients at imminent risk of death as these conversations are then focused only on eliciting resuscitative preferences [81]. By the time a patient becomes critically ill and requires ICU admission, this may also be too late [82, 83]. High-risk hospitalized patients who are not so distressed from their acute illness and are not ready for hospital discharge most likely represent the ideal targets for GOCDs. An initial approach at 1 to 2 days after admission may represent the period of hospitalization associated with sufficient stabilization to permit fulsome discussions with family and substitute decision-makers. One timely visit should be sufficient in most cases as the value of allowing patients, family, and substitute decision-makers to deliberate about their care preferences may not add value to their final treatment decisions [84] and may be impractical from the perspective of time and clinical workflow. All modifications to the protocol which may impact on the conduct of the study and potential benefit of the participant or may affect participant safety, including changes of study objectives, study design, participant population, sample sizes, study procedures, or significant administrative aspects, will require a formal amendment to the protocol. Such amendments will be agreed upon by the principal investigators, and the RVH Research Institute, and approved by the RVH Research Ethics Board prior to implementation. Administrative changes of the protocol are defined as minor corrections and/or clarifications that have no effect on the way the study is to be conducted. These administrative changes will be agreed upon by the principal investigators, and the RVH Research Institute, and will be documented in a Note to File to the RVH Research Ethics Board. Prior to the initiation of the study at each study site, the RVH Research Institute will be responsible for providing adequate training to the healthcare providers and study personnel. The training will cover all aspects of the study protocol and procedures and will include practical training on the use of the randomization system, electronic case report forms, and study materials such as the f-GOCD ASKmeGOC e-tool. Written and electronic materials will be supplied for study personnel and for the education of the healthcare providers. An independent study monitor from the RVH Research Institute will ensure that the study is conducted according to the protocol, good clinic practice guidelines, and relevant regulatory requirements. The main duty of the study monitor is to help the principal investigators and the RVH Research Institute maintain a high level of ethical, scientific, technical, and regulatory quality throughout all aspects of the trial. The principal investigators, healthcare providers, and study personnel will assist the study monitor by providing all appropriate documentation and being available to discuss the study. At the completion of the trial, a final monitoring and close-out visit will be conducted by the study monitor. Supplementary Information Additional file 1. SPIRIT 2013 Checklist: Recommended items to address in a clinical trial protocol and related documents. Acknowledgements We would like to thank the Royal Victoria Regional Health Centre Research Institute (Dr. Jesse McLean, director), and all their staff for their support throughout the study development, and the Georgian College Computer Studies (Dr. Rich Freeman, professor) program for their collaboration in creating a web-based version of the ASKmeGOC e-tool. Trial status Protocol version 3.0 (April 17, 2022) Recruitment start date October 3, 2023 Recruitment completion data (expected) February 28, 2026 Abbreviations Advanced care plans Consolidated Framework for Implementation Research Confidence intervals Cardiopulmonary resuscitation Clinical Research Associate Do not attempt cardiopulmonary resuscitation Electronic medical record Goals of care discussions Facilitated goals of care discussions Healthcare provider Intensive care unit Life-sustaining therapies Personal Health Information Protection Act Authors’ contributions GD developed the initial ASKmeGOC e-tool. GD, CM, and DA revised the ASKmeGOC e-tool used in this study. GD, CM, DA, MA, and KC contributed to the conception of this study. GD, CM, DA, MA, and KC contributed to the writing, review, and editing of this protocol. GD, CM, DA, MA, and KC provided final approval of this protocol. GD, CM, DA, MA, and KC are accountable for all aspects of the study. GD also secured study funding, developed all study materials and data collection instruments, and prepared the statistical analysis plan. Funding Funding for the study was provided through an unrestricted grant from the Alectra Fund for Health and Social Innovation administered through the Royal Victoria Regional Health Centre Foundation. The funders had no role in the study design, data collection, data analysis, manuscript preparation, or publication decisions. Data availability All de-identified study data will be made available upon request. Declarations Ethics approval and consent to participate The study received ethics approval by the Royal Victoria Regional Health Centre Research Ethics Board (September 2, 2022; REB Study No. R22-003). For patients randomized to the f-GOCD group, study ICU nurses will obtain verbal consent from competent patients or, otherwise, their substitute decision-makers. A waiver of the requirements for signed informed consent is requested for the f-GOCD intervention arm of the study. This approach is justified in comparative effectiveness trials involving the randomization of patients to one or more standard of care interventions [85] because the trial poses no more than minimal risk, the alteration from written to verbal consent does not materially affect the rights or welfare of the patient, and verbal consent would be the usual approach used to initiate any GOCD with any patient and/or their substitute decision-maker during routine clinical care. In our pilot study, we used verbal consent from patients or their substitute decision-makers without any adverse consequences documented. Using this approach, 90.2% (95% CI 76.9, 97.3) of patients or their substitute decision-makers consented to receiving the f-GOCD intervention (42). In that study, 100% of patients or their substitute decision-makers completed the intervention. In addition, these patients will be approached for written verbal consent, documented by electronic e-signatures via REDCap by the Clinical Research Associate (CRA) after completion of the GOCD intervention and asked to complete the 7-item SDM GOCD tool evaluation questionnaire. For those participants who consent, the CRA may help them complete the survey during the index hospitalization. These surveys will only be completed during the index hospitalization. Only completed surveys will be compared between groups. For all participants in both the control and intervention arms, the CRA will seek verbal consent to complete the Quality of Communications and 4-item SURE questionnaires. These questionnaires will only be available for completion during the index hospitalization. The CRA will approach these patients for verbal consent after their CODE STATUS has been completed and documented in the EMR. The rationale for not requiring written consent in this part of the study is a quality improvement initiative as the results of the two questionnaires will be fed back to the physicians who signed the patient’s CODE STATUS and the organization so that subsequent improvements can be made in how we approach patients and their substitute decision-makers for end-of-life care decisions. Patients or their substitute decision-makers will be told that they may refuse to participate or withdraw their participation after consent at any time without fear of their ongoing care being compromised. Patients or their substitute decision-makers will be informed that any preferences documented in the CODE STATUS can be changed at any time during their hospitalization, and the trial personnel or principal investigators will be available to answer any questions they may have after the f-GOCD intervention has been completed. Trial personnel will be available to review any questions related to a previously completed CODE STATUS and will also be available to assist patients or their substitute decision-makers in revising these at any time during their index hospitalization. Consent for publication Not applicable. Competing interests The authors declare that they have no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Additional file 1. SPIRIT 2013 Checklist: Recommended items to address in a clinical trial protocol and related documents. Data Availability Statement All de-identified study data will be made available upon request.