Phase 1: competitive analysis
We identified 16 available DAs related to RRT decision making through a systematic environmental scan. A proposed summary of the resources’ features and content elements is presented in Additional file 3. Most resources were from USA (N = 7), and the rest were developed in UK, Canada, Singapore, Australia, Denmark, Spain, and Sweden. Currently, the DAs designed for RRT decision making mainly cater to chronically ill patients: ‘End of Life Issues (N = 2)’ the advance care planning of kidney failure; ‘End Stage Renal Disease (N = 5)’ the choices of transplantation, dialysis or conservative management; ‘Chronic Kidney Disease (N = 9)’ dialysis modality options. The included studies showed that there is no DA on the topic of RRT decision making has been established for critical illness. Therefore, the evidence-based necessity to develop a target one for surrogates of critically ill patients requiring RRT was confirmed.
The two on end-of-life issues [44, 45], which highly overlapped with our target goals, only provided users with general information (e.g., you may feel better physically, you may be able to return to normal activities, you will no longer have side effects or problems from dialysis, your quality of life may be better, etc.), while details about how these outcomes would otherwise impact patients or about which factors trigger which outcomes were not available. Given the expectation of prognosis between ICU physicians and surrogate decision-makers was discordant [65], our potential DA should better illuminate the effects of clinical outcomes and, further, explain what outcome could be best achieved by prudently and deliberately decision making during life-saving moments. Pros and cons of each choice carried out on a contrast table were recommended to solve this problem. It is important to present relevant potential outcomes as much as possible before decision makers made the final life-saving decision, which can mostly prevent decisional conflicts or regrets.
There is not one single type of sources to develop a DA. Qualitative interview (i.e., individual interview or focus group) combined with systematic literature review was the most selective source of developmental data [46–49]. Other developmental sources including the construction of risk prediction model by national surveillance data [50], national and international patient needs through online survey [51, 52], educational material [53, 54], previous decision-making experience [53], and patient flow mapping [54]. The source of DA development depends on the characteristics of population coverage, for example, whether the scope of the population is specific or general, or whether the type of decision making is user preference-sensitive or not. For our targeted users, surrogate decision makers of critically ill patients under Chinese medical decision-making background, it is very important to collect their decisional needs and experience based on this special context. Qualitative interview could be the effective tool to provide us with preliminary sources.
For the display formats, two-thirds were web-based, while others combined with mobile applications, videos, PDF format or paper leaflets. There is no doubt that digital DAs are by far dominating the development market. Interactive features might be the possible reason for the boom. As if digital DAs were available for content control, timely feedback, social support, and other interactive features, users’ decision-making experiences could be optimized further. Given major modifications could be easily made to the prototype by digital technologies, it might also be the best way to provide iterative refinement for DA development. Obviously, it is necessary to apply digital technology to develop a DA with user-centered design.
Regarding the methodology of development, seven (7/10) employed empirical and theoretical frameworks [47–49, 51, 55–57], whereas six (6/16) did not report their development protocols [53, 54, 58–61]. The methodologies applied in their development process including the IPDAS criteria [62], the principles of shared decision-making process [63], Option Grid [64], the systematic development process for patient decision aids [24], and other clinical or service guidelines. In addition, some were also inspired by earlier feasible DAs [49]. However, only one study reported a systematic protocol conceived before development [50]. Note that we have previously published a study protocol for the development [28], and we would followed with the guidance of the IPDAS criteria [62].
Phase 2: User Needs Assessment
We invited 15 family surrogate decision-makers to participate in decisional needs assessment. The median (interquartile range [IQR]) age of surrogates was 48 (39.5–56.0) years. The mean APACHE II score for patients was 26 ± 7.9 scores. The mean length of stay in the ICU was ten days (ranging from 3d-28d). The demographic characteristics of family surrogates and their loved ones are presented in Table 1, and the qualitative scripts of family surrogates’ perceptions during RRT decision making is presented in Additional file 4.
We identified four themes with descriptive analyses of surrogates’ decision-making experience (Fig. 2). First, from an overall perspective, family surrogates stuck into decisional dilemmas after experiencing perplexity during RRT decision making. Most participants, even well-educated, could not make reasonable judgements with the information available to them due to the lack of specific guidance. Although participating in shared decision making with clinicians, some surrogates still deemed RRT selections as complex requirements. In addition, there were no criteria or guidelines suggesting the best suitable option for them before the decision making was required by clinicians, which further intensified these dilemmas. Second, family surrogates usually experienced uncertainty during the decision-making process. They reflected that they would be much more decisive if they could be informed about the ratio of risks and benefits for choosing RRT, while in reality they were only provided with rough guesses by clinicians. The unpredictable outcomes of each option likewise induced a sense of uncertainty and anxiety.
Third, surrogates’ capacity for decision making was constrained. Most comments from respondents reflected a clinician-oriented model in the decision-making process. Their described feeling as though they were being informed of the patients’ disease and treatment plan, but that the dominant deciding role was occupied by the doctor. Deficiencies of time and opportunity was another restriction. Surrogates reflected that, when the RRT decision making happened suddenly in the ICU, there was no time to think anything other than that their loved one might die without treatment. Furthermore, the self-efficacy of surrogate decision makers was expressed as insufficient. Most of them were unable to clarify the preferences and values of the patients when the decision making happened, while others were incapable of determining the trade-off between the benefits and the risks of treatment. Finally, surrogates responded to RRT decision making with delayed confirmation. When the decision about whether to initiate RRT for critically ill patients was confirmed by clinicians after acquiring the informed consent from surrogates, surrogates would reconstruct their decision later, through information searching or by seeking peer support to obtain familiar and trusted resources in hope of resolve their confusion – often to no avail.
Given the guidance of Ottawa Decision Support Framework [31], surrogates’ decisional needs were identified further, and the specific decision support was constructed (Fig. 2), based on the above four domains of surrogates’ perspectives regarding RRT decision-making experiences in ICU settings. In terms of surrogates’ decisional dilemmas and the sense of uncertainty, decisional conflict was the main barrier for them to further shared decision making with professionals, as available resources were unable to support surrogate decision making in ICU. Therefore, the information needs were identified, and the potential DA need to provide sufficient information for RRT choices clarification, facts and probabilities. For the limited decisional capacity, surrogate decision type (clinician-oriented model), timing (deficiency), stage (low self-efficacy), leaning (unable to clarify preferences) were presented. It is significant to provide additional professional support and values clarification besides DA intervention during surrogate decision making process, through clarifying their needs and preferences in specific decision-making context in reality. Regarding surrogates’ delayed confirmation in RRT decision making by self-searching information and seeking for peer support after the decision making was made, their expectations on knowledge, resources and other support were identified. Hence, corrective role guidance of surrogate decision maker should be better integrated into the DA to guide deliberation and communication. These measures indicated the main targets we needed to account for in DA development.
Table 1
The demographic characteristics of family surrogates and their loved ones (N = 15, respectively)
Participant characteristic | Patients, n (%) | Surrogates, n (%) |
Age, y | | |
Mean, SD | 52 (20.3) | 49 (12.1) |
Median (interquartile range) | 57 (35.5–66.5) | 48 (39.5–56.0) |
< 40 | 4 (27) | 4 (27) |
40–59 | 5 (33) | 8 (53) |
≥60 | 6 (40) | 3 (20) |
Gender (man) | 11 (73) | 12 (80) |
Reasons for ICU admissiona | | |
Nephrogenic | 8 (53) | - |
Pulmonary infection | 7 (47) | - |
Respiratory failure | 6 (40) | - |
Cardiovascular | 6 (40) | - |
Sepsis | 3 (20) | - |
Gastrointestinal | 3 (20) | - |
Neurologic | 2 (13) | - |
Metabolic | 1 (7) | - |
Trauma | 1 (7) | - |
APACHE II score (Mean, SD) | 26 (7.9) | - |
ICU-LOS (Mean, min-max) | 10 (3–28) | - |
Reasons for ICU discharge | | |
Financial burden/ forgo therapy | 8 (53) | - |
Transfer to the general ward | 6 (40) | - |
Death | 1 (7) | - |
Relationship to patient | | |
Children | - | 5 (33) |
Parents | - | 4 (27) |
Spouse | - | 3 (20) |
Other | - | 3 (20) |
Educational level | | |
College/university | - | 9 (60) |
High school | - | 4 (27) |
Junior middle school and below | - | 2 (13) |
a More than one choice possible. APACHE: Acute Physiology and Chronic Health Evaluation. ICU-LOS: length of stay in ICU.
Phase 3: User Persona
We recruited 15 family surrogates with RRT decision-making experience in the ICU to produce a user persona, which described the typical user expected to engage with the DA and informed subsequent design decisions. The demographic characteristics of participants is presented in Table 1. Since most of the surrogate decision makers in ICU are men, we created ‘Bobby’, our user persona. He is a family surrogate decision maker for his elderly father, who has been diagnosed with septic AKI after ICU admission.
Bobby’s primary goal is to decide whether or not to choose RRT based on information about his father’s disease’s progression and available treatment options while his father is in a coma and incapable of decision making. To achieve this goal, he expects to be provided with sufficient information about the available choices and guided by a professional who will ask him how to proceed at each step. He is confused when communicating with ICU clinicians, because there is indeed a knowledge gap between himself and the doctors which renders him unable to engage in a successful discussion with the medical professionals. He is particularly perplexed by ambiguous information about the treatment outcomes and finds it impossible to weigh the potential risks and benefits of RRT without numeric estimations. When participating in the decision making, Bobby is frustrated by uncontrolled factors related to his father’s disease and treatment progress, especially if negative information about the procedure is not mentioned before the decision is made.
Bobby expects a DA can facilitate his interaction with clinicians, especially when he suffers psychological distress because of his father’s ICU admission. He wants plain explanations to clarify this complex choice so that he can prepare to participate in decision making. The brief and concise content in the DA will be better suited to his needs so that he can understand the situation as quickly as possible, while additional detailed information should be available as a supplement. A digital format will be the best mode of presentation for the potential DA, because he thinks this is most convenient readily available, while paper pamphlets can be prepared if they are needed in different settings.
He sees himself as passive and feels a total mess in his mind when anticipates writing an informed consent for RRT for his father. He thinks it is necessary to take a role in choosing treatment for his father, but he worries about his ability to make the best choice. He is concerned that he may have future regrets if he makes an incorrect decision.
Bobby plans to increase his knowledge about his father’s disease and the option of RRT by self-searching on the internet (usually acquiring free results and sample phrase segments through online search engine) or by seeking peer support from others with similar experiences, if possible, but admits he feels stuck within a decision dilemma to some extent.
Phase 4: Evidence Synthesis
Given the preference-sensitive nature and ethical conflicts the choice of RRT presents to the surrogates of critically ill patients, we conducted a separate process to check the relevance of and to update, where needed, 1) the introduction of critical illness with AKI and the RRT, 2) the benefits and risks of RRT selection, 3) the ICU and post-ICU outcomes of critically ill patients receiving RRT as it currently exists.
Evidence searching occurred between Dec 2019 and May 2020. A total of 15, 220 records were identified from database and website searches. In total, 27 studies were identified as suitable for inclusion, where Fig. 3 presents the screening process and results. The characteristics of the included literatures are summarized in Additional file 5. Of the 27 included articles, five were published before 2015 and the remaining selection were published in 2015 (n = 1), 2017 (n = 3), 2018 (n = 2), 2019 (n = 6), and 2020 (n = 7). All the articles were published in English except for four published in Chinese. Regarding the types of publication, four were the highest quality of evidence available from UpToDate, while the remaining were published as clinical guideline (n = 6), expert consensus (n = 3), systematic reviews or meta-analyses (n = 10), and original studies (n = 4).
The quality evaluation of guidelines appraised by AGREE II, shows that two were recommend at Level A and four at Level B. All systematic reviews and meta-analyses were evaluated as high quality except for two, which were evaluated as middle quality due to incomplete consideration of the evidence-based method. Two expert opinions were evaluated as high quality and one as middle quality because of the unclear consistency when compared with other studies. Concerning the original studies, two non-randomized controlled trials were evaluated as middle quality since no measure taken when the clinical follow-up lost; the cross-sectional study was evaluated as high quality; the prospective cohort research was deemed as middle quality because three items were evaluated with unknown/unclear, including measures to solve confounding factors, initial observational outcomes, and measures to solve incomplete follow-up.
A total of 21 evidence items were extracted from the above articles through in-depth reading, analysis, and summarization. Based on the framework from the results of needs assessment and the design guidance offered by the user persona, all decision supports available from UpToDate, and relevant guidelines were integrated into background information detailing a general introduction of the disease and the treatment. An additional 21 items of evidence (see Additional file 6) were cataloged detailing the benefits and risks of RRT, possible outcomes, and the reasons to choose this treatment.
Draft Prototype
We drafted the prototype according to the elements from competitive analysis (design tactics and strategies), user needs assessment (targeted users’ unmet decisional needs and potential decision support), user persona (context immersion for development), and evidence synthesis (latest clinical evidence of the treatment decision making for critically ill patients requiring RRT). A crucial step for prototype drafting was to reach a consensus within our project team, especially, with respect to expert recommendations on evidence transformation and presentation. The DA was developed with two main parts, a “user guide” which refers to a user instruction to state decision-making problems on patient right and interests, and a “decision guide” to lead the shared decision-making process through a systematical guidance. Specifically, the DA was rapidly prototyped with simplified navigation interface, and with interactive features such as content control, tailored decision-making logic, and timely feedback. A conjunction of contrast table, smiley matrix, pie chart, curve chart, bar graph, etc., was comprehensively applied to transform the numerical data into user views. Finally, the digital DA was developed through FAISCO [66], a self-service website construction system in China. The content framework and key features of the prototype are presented in Fig. 4, and final DA prototype is provided in Additional file 7.