Shared decision-making (SDM) is a model of patient-centered care that encourages patients and clinicians to work together to reach medical decisions by weighing the risks and benefits of various options within the context of the values and goals of the patient [1]. SDM is the process of communication, deliberation, and decision-making between clinicians and their patients [2,3]. During this process, treatment options, including risks and benefits, are discussed and patient preferences are explored to inform decision-making. Previous research has shown that SDM increases patient engagement, knowledge, risk comprehension, and participation in decision-making [4-6]. Furthermore, SDM has been shown to improve patient–health care provider communication, patient satisfaction, compliance, and clinical outcomes [7,8]. Notably, the US Preventive Services Task Force considers SDM an ethical right of all patients, independent of health outcomes [9].

Currently, SDM is not routinely required by large health care payers, such as the Centers for Medicare and Medicaid Services, which is a US federal agency providing health care for patients with low income or older adult patients. Medicare is the largest single payer for health care in the United States [10]. The Centers for Medicare and Medicaid Services is slowly adopting SDM standards for limited and specific procedures, conditions, and devices, including lung cancer screening, atrial fibrillation, and implantable defibrillators [11]. US states such as Washington and Vermont are also beginning to mandate SDM for specific conditions and other states such as Connecticut, Massachusetts, Maine, Minnesota, New Hampshire, and Oklahoma are considering legislation [12].

The Health Information Technology for Economic and Clinical Health Act was adopted in the United States in 2009, enabling the widespread adoption of electronic health record (EHR) systems across the US health care system [13]. Leveraging EHR systems to promote the use and uptake of SDM appears to be a logical approach to operationalizing SDM programs. For example, Kuo et al [7] found several positive effects of integrating SDM tools within an EHR system, which include improved clinical outcomes, positive lifestyle behavior changes, more deliberation with clinicians, and less decisional conflict. Despite the interest in incorporating SDM into routine care, current research studies identify a variety of obstacles that limit SDM adoption. Some of these obstacles include technical integration issues; logistical and workflow challenges; and psychological impediments, such as uncertainty and legacy belief systems, which continue to impede progress [14-16]. Integrating SDM tools and processes into EHR systems is often a complex and difficult problem [17-21].

SDM may involve the use of a decision aid, a type of tool that helps patients consider options for medical decision-making by increasing patient knowledge; illuminating treatment options including risks, benefits, and efficacy; and exploring patient preferences and values [22-24]. The Ottawa Decision Support Framework provides recommendations and guidelines for health care decisions and standardized checklists for the development and evaluation of decision aids based on the International Patient Decision Aid Standards [25]. SDM components found within the International Patient Decision Aid Standards include clarifying patient values, coaching and guiding patients when making decisions, and ensuring literacy levels are addressed [26-28]. Decision aids developed using the Ottawa Decision Support Framework as well as other decision aids outperformed the usual care for performance in a large overview of systematic reviews [25,28]. Patient decision aids can play a key role in facilitating the occurrence of SDM at the point of care as clinicians may use a decision aid to support SDM conversations. Furthermore, decision aids may provide a way to standardize SDM experiences. Notably, decision aid research developed before the widespread adoption of EHRs [29-32]. Decision aids are not synonymous with SDM processes as SDM processes may incorporate the use of a decision aid or may use a conversation between patient and clinician without the use of a decision aid. Decision aids typically contain patient education as part of the decision aid content; however, a clinician may choose to verbally educate a patient without the use of a decision aid. Previous research has shown that decision aids integrated into an EHR system show improvement in patient-centered outcomes, including a reduction in decisional conflict. However, there are significant barriers and complexities to overcome when integrating them into commercial EHR systems [33-35].

Both decision aids and SDM tools still face substantial hurdles to clinician uptake [15,32,36-38]. Gaining the attention of health care providers can be difficult as SDM relies on health care provider discretion, time, and required effort [15,39-41]. Integration into the EHR system is recognized as an important strategy as it enables SDM to be part of clinical workflow [42]. However, integration can also pose challenges because unavoidable health care provider alerts run the risk of high levels of alert fatigue or reactance. SDM tool developers and health care systems must be judicious in their choice of tools to develop and integrate. It is important to understand the characteristics of SDM tools that have been integrated into EHR systems.

A literature review revealed limited studies describing the integration and implementation of SDM tools within an EHR workflow [43-46]. A number of recent systematic reviews have been published on the general topic of SDM [19,33,47-58]. These studies focused on a range of SDM topics, including SDM within specific surgical specialties, behavioral theories underpinning SDM, organizational characteristics to support implementation, validated SDM measurement tools, SDM key components, and quality of SDM. However, at the time of the protocol registration, we were unable to identify a systematic or scoping review that specifically addressed the characteristics of SDM tools that have been integrated into an EHR. Thus, this study aimed to understand the characteristics of SDM tools that have been integrated into an EHR system and implemented within a health care system by conducting a scoping review.

This scoping review followed methods described by Arksey and O’Malley [59], Levac et al [60], and the Joanna Briggs Institute (JBI) [61]. The study protocol is published with the Protocol Registry of Evidence Reviews at the University of Utah. Consensus was achieved on the inclusion and exclusion criteria and primary research question. For transparency and reproducibility, we followed the PRISMA-ScR (Preferred Reporting Items For Systematic Reviews And Meta-Analyses Extension For Scoping Reviews) and PRISMA-S (Preferred Reporting Items For Systematic Reviews And Meta-Analyses Search Extension) reporting guidelines for the scoping review protocol and manuscript [62,63].

An information specialist developed the search strategies using database-specific subject headings, keywords, and team feedback for the key concepts of SDM, patient engagement, decision support, and EHRs. Example publications were provided to the information specialist for strategy development and term harvesting. The search strategy was evaluated by a librarian coauthor using the Peer Review of Electronic Search Strategies guidelines [64]. EndNote (version 20; Clarivate) was used to manage citations and remove duplicates. Covidence (Veritas Health Innovation) provided a second pass for duplicate removal. Search strategies are provided in Multimedia Appendix 1.

The following databases were used to identify eligible studies between 2009 and 2021: MEDLINE (via Ovid platform), Embase (via Embase platform), CINAHL Complete (via EBSCOhost platform), Cochrane Library (via Wiley platform) including CENTRAL (via Wiley platform), PsycINFO (via EBSCOhost platform), Scopus (via Scopus platform), and Web of Science Core Collection (via Clarivate Analytics). No filters other than date limits were applied. The date limit from 2009 corresponds to the adoption of the Health Information Technology for Economic and Clinical Health Act in the United States [65].

Covidence software was used for screening. Four researchers independently screened titles and abstracts for inclusion and exclusion criteria and met for discussion. The protocol was modified during this phase to expand inclusion criteria based on discussions among the screeners. For example, after finding a number of design studies and concept papers, we added the requirement that all studies must have SDM tools that have been implemented into a real-world clinical setting. Furthermore, we excluded retrospective studies after identifying studies that focused on a retrospective analysis of clinical notes and activities. Changes to the protocol are listed in Textbox 1. Following the screening process, 303 (10%) of the 3017 articles were identified for full eligibility assessment.

Expanded screening criteria were added to the original inclusion and exclusion criteria identified in the research protocol, as noted in Textbox 1. The additional categories were identified and added as the team recognized additional screening criteria as screening commenced.

A team of 5 researchers, including 2 professors with extensive backgrounds in cognitive and sociotechnical informatics (CW and JMB), 1 informatics researcher experienced in sociotechnical design (TT), an informatics student (JHP), and another informatics student (WR II), met on a weekly basis to discuss inclusion and exclusion criteria and extraction goals for research studies. Initially, each researcher was assigned the same 5 studies to assess for eligibility review. Then, the research team discussed and deliberated on whether each study met the established inclusion criteria until the team achieved consensus. A spreadsheet was created with each evaluation component, including basic study characteristics, integration, and implementation methods. Study characteristics included study type, SDM intervention, clinical domain and conditions, measures, outcomes, integration approaches, and implementation processes. The research team discussed the criteria and achieved consensus for spreadsheet components through discussion, debate, and review. The reviewers then evaluated approximately 10 studies per week as a group, using the spreadsheet to determine criteria for analysis against the study aims. After this, the remaining studies were distributed among the team members for a full review of eligibility. We excluded studies that did not report measurable quantitative outcomes, were not integrated into an EHR system, did not involve a shared decision (with or without a decision aid), and were not in person at the point of care. EHR integration for the purpose of this review ranged from scanned PDFs appended to EHR patient records to full technical integration using semantic interoperability methods.

We identified the studies of tools designed to support SDM that were integrated into the EHR system for evaluation and reported a total of 18 studies, which were included for final review. See PRISMA-ScR Figure 1 and Multimedia Appendix 2.

This scoping review included 18 studies that met the inclusion criteria (Figure 1), including 15 (83%) quantitative and 3 (17%) mixed methods studies. The most commonly used SDM outcome measures are listed in Table 1.

Of the 18 studies, 7 (39%) measured SDM components, such as patient empowerment, self-efficacy, self-determination, and goal setting. The 7 (39%) studies measuring use rates reported various levels of use; however, predetermined targeted use rate goals were not noted. We found that at least 3 (17%) of the studies achieved a use rate of >50% of the targeted users. In total, 4 (80%) of the 5 studies using clinical outcome measures reported positive improvements based on the SDM intervention [71,72]. Of the 18 studies, 4 (22%) studies that measured patient satisfaction included elements of communication quality, such as helpfulness, respect for the patient’s opinion, understandability, confidence in the health care team, and worry reduction [72-75]. Documentation rates increased for all 4 (22%) studies using documentation as the primary outcome measure [76-79]. A total of 2 (11%) studies that measured clinician satisfaction reported clinician satisfaction at >85% and preference for a systematic approach via a tool at 100% [71,80]. In total, 4 (22%) included studies incorporated patient goals and values in the SDM process [72,78,81,82]. Consistent with previous research, 1 (5%) study reported that obstacles to tool use included a lack of awareness of the tool and low confidence in how to use the tool [83].

We categorized the included studies by macrolevel functional goals to include (1) care planning and goal setting (10/18, 56% studies), (2) prevention and screening (4/18, 22% studies), and (3) medical management and treatment (4/18, 22% studies; Table 2). The largest percentage of the care planning and goal setting category was end-of-life planning [76,77,81,82], followed by communications [72,84]. In total, 75% (3/4) of the prevention and screening studies focused on the detection of cardiovascular disease [71,74,85], and 75% (3/4) of the medication management and treatment decision-making studies focused on diabetes and related pharmacotherapy [80,83,86].

Tool objectives were generally a direct function of the use of the SDM tool; for example, an end-of-life planning template may have a stated objective to increase documentation rates for communications. Patient satisfaction–related outcomes, such as decision satisfaction because of a shared conversation, were not noted in 11 (61%) of the 18 studies [72,73,75,80-84,86-88].

This review summarizes the findings of 18 studies that investigated approaches to the integration and implementation of SDM tools within EHR systems. Prominent aspects found across studies included functional SDM goals, such as communication, documentation of communication and discussions, and SDM template use. We identified only 4 (22%) randomized controlled trials (RCTs), which may be a reflection of the operational nature of the study measures. In 10 (56%) studies, clinicians were the primary targeted users, whereas 4 (22%) studies targeted patients and 5 (28%) targeted both. Integrating SDM interventions into a clinician’s workflow within an EHR system appears to support improved SDM uptake [60,89-91]. Studies reported multiple EHR and workflow integration approaches, which may be an indication that a dynamic model that is scalable could help guide the adoption of SDM tools and processes across different environments. Future research to understand the utility of adjusting SDM model depth and complexity based on the magnitude of the decision, the complexity of the clinical situation, and the patient preferences is warranted.

Most (10/18, 56%) of the included studies focused on SDM tools that were functionally designed to support care planning and goal setting. We also noted 1 (6%) study that used the concept of open medical notes as a precursor to shared care conversations.

Most (4/18, 22%) prevention and screening studies targeted by SDM focused on common conditions, such as the detection of heart disease, colorectal cancer, and other common conditions. Future opportunities for EHR-integrated SDM screening tools include the assessment of social determinants of health for the purpose of supporting individuals with social needs. Furthermore, mental health SDM screening tools offer an important area for prevention and screening to detect anxiety, depression, loneliness, addiction, and suicide risk, as a few examples.

Although studies generally lacked sufficient technical detail on the actual integration method into each EHR system, we found considerable variation in how SDM tools are implemented within EHR workflows. The implementation approaches for SDM applications showed varying levels of planning and efforts to promote SDM intervention awareness. For example, 33% (6/18) of the studies did not describe implementation activities for clinicians, whereas other (6/18, 33%) studies described more detailed implementation plans that included training, demonstrations, knowledge queries, and incentives [71,73,77,78,81,82].

In total, 22% (4/18) of the studies included in the review suggested that integrating an SDM intervention into a clinician’s EHR workflow may support increased awareness of the intervention [60,72,89,90]. This represents an important step toward supporting clinicians in the adoption and use of SDM tools and processes. Creating awareness, interest, and commitment to SDM tool use is often overlooked and may contribute to waning adoption for otherwise well-designed SDM tools, potentially leading to adoption failure. The implementation plan is a critical component to moving people and the organizational culture toward the adoption of SDM tools. These steps also provide a forum for direct feedback should clinicians lack confidence in the accuracy or efficacy of an SDM tool, as was noted by Ballard et al [83].

This review found that established processes and required patient interactions, such as end-of-life care planning protocols, appear to support greater SDM adoption success because they rely on workflows and processes that are already in practice. Adoption through incremental change, such as an evolutionary approach rather than a revolution, may be more likely to gain momentum over time [85].

Our review found that only a few randomized clinical studies were noted. Future research studies should assess the effect of SDM in multisite pragmatic RCTs.

This review identified a wide range of approaches to how SDM interventions are organized and delivered within the clinical setting. Future research could explore contextually aware tiered levels of intervention. This is important to standardize SDM care processes within specialties in order to support large-scale adoption of validated SDM tools and methods while addressing practical considerations within the constraints of each specialty.

Our review found that 17% (3/18) of the studies attempted to engage patients before a clinical encounter; however, no studies attempted to engage patients in decision-making after a clinical encounter [76,79,84]. Protracted decisions that include a series of small decisions align well with the use of patient portals to begin preparing patients for the SDM process. Another area for future research could be to understand the efficacy of postclinical encounter engagement to address the ongoing maintenance and reinforcement of decisions that have been made in collaboration with the clinician.

We found limited studies that included the identification and measurement of patient values. Thus, patient involvement in establishing their values and alignment of care with those values appear to be areas for additional focus in future research. Involving patients in goal setting is a key element to support patient engagement, which is essential to help patients manage their decisions outside of the clinician encounter [92].

This scoping review revealed a wide range of approaches to the integration and implementation of SDM tools within an EHR workflow. We found that SDM tool integration methods ranged from simple scan and save methods to fully integrated and operable technical integration with an EHR system. Moreover, implementation approaches were highly variable, ranging from sending an email to clinicians to create awareness to more complete programs involving demonstrations, training, incentives, clinician-led promotion, and feedback sessions. Areas for future research include more RCTs, the development of contextually aware SDM interventions, longitudinal patient engagement with pre- and postvisit interventions, the study of patient-driven goals and values, and the subsequent impact on patient satisfaction levels. Future studies should detail pragmatic and actionable integration and implementation steps to advance the uptake of SDM.