There is increasing recognition of the ethical imperative to support patients to be engaged in their care, especially in health-related decisions. Shared decision-making (SDM) is a collaborative process whereby health care professionals support patients in making decisions that are informed by the best evidence and by what matters to them [1]. SDM improves the health care experiences of patients and health care professionals and leads to better health care processes, patient outcomes, and lower health costs [2-5]. SDM is the best practice for informed consent and is fundamental to patient- and family-centered care [6]. However, SDM has not yet been widely implemented in clinical practice because of several perceived barriers [7]. In some contexts, SDM implementation is encouraged by health policies, but certain challenges related to patients, patient–health care professional relationships, and organizational factors remain in the way of its concrete adoption [8].

Training programs for health care professionals [9,10] are believed to be crucial to the implementation of SDM. However, SDM training programs for health care professionals are highly heterogeneous [10], and we still do not know what makes them effective [9]. One promising approach is reflexivity, which has been shown to increase health care professionals’ willingness to be more engaged in the health care offer and collaborate with other professionals [11]. SDM training programs that use reflexivity strategies have the potential to be more effective than those that do not [12].

Reflexivity is a form of learning based on reflection on one’s experiences applied in a professional or interprofessional context [13]. Concepts related to reflexivity are found in various disciplinary fields under different names. Writing about reflective practice (a practice based on reflexivity), Schon [14] states that the concept entails critical thinking concerning the actions and stances one takes [14,15]. In health care practices, being reflexive can mean different things: acknowledging and questioning the power dynamics implicit in a health care encounter; identifying the assumptions that underlie a health care situation; or examining the influences, such as values and beliefs, that shape health care practices [13].

On the basis of this multiplicity of definitions, Sandars [16] developed a guide that classifies them into 3 main approaches. According to Sandars’s work [16], reflection is a form of learning that is based on three common aims: (1) reflection for learning, (2) reflection to develop a therapeutic relationship, and (3) reflection to develop professional practice [14,15]. Indeed, reflexivity includes questioning the premises of an action, such as the values, norms, and beliefs that a professional may hold, as well as how such actors justify their actions [17]. In sum, reflexivity is a good strategy for motivating both health care professionals and patients to engage in patient care in a collaborative way.

Reflexivity is appropriate in the context of SDM because the latter emphasizes a partnership between patients and clinicians in making decisions and establishing care plans. SDM aims to reposition the knowledge of patients and clinicians on an equal footing, adjusting the asymmetrical power relationship between patients and health care professionals [18,19]. A prerequisite for implementing SDM in care settings is that health care professionals not only have the knowledge and skills but also the willingness to engage patients in the decision-making process [20].

Training based on reflexivity may lead health care professionals to question the power issues inherent in a more traditional conception of the health system and of their role and may motivate them to adopt SDM [11]. Thus, we hypothesized that SDM training programs that integrate reflexivity strategies would be more effective in increasing the adoption of SDM than those that do not.

Research also shows that SDM training programs developed with an interprofessional orientation are to be encouraged [21]. Interprofessionality is defined by D’amour et al [22] as the development of a cohesive practice between professionals from different disciplines for the care of a single patient. Specifically, interprofessional collaboration involves collegial, authentic, constructive, open and honest communication as well as mutual trust and respect between professionals who are committed to achieving a common set of goals. However, although an interprofessional orientation is highly encouraged in SDM because of the need for better teamwork and professionals’ openness to other forms of practices, few training sessions with an interprofessional orientation are available [10]. In addition, there is little evidence that interprofessionally oriented training is more effective than other training approaches. Given the similarity between the methods used in training programs with an interprofessional orientation and those used in training programs with certain reflexivity strategies (eg, peer-to-peer group learning), it is important to examine how effective both these training types are.

In the Cochrane review, one of the criteria for selecting programs was that the type of measurement followed must be a patient-reported outcome measurement (PROM) or an observer-reported outcome measurement (OBOM). A PROM is an instrument used to collect information directly from patients. PROMs do not require amendments or interpretation by a clinician or another observer [23]. An OBOM is any instrument used by a third-party observer to report observable concepts such as signs or behaviors to assess, for example, the decision-making process during an encounter between a patient and family and their health care professional when facing a health treatment or screening decisions [23]. It seemed useful to see what types of measurements were most common, especially among the effective training programs that used reflexivity strategies. This could offer another potential reason for a training program’s effectiveness. As we were interested in interventions targeting health care professionals, it was important to discern whether the type of measurement used to evaluate the intervention was equipped to examine the different levels of effectiveness as defined by Kirkpatrick, the creator of one of the most common evaluation models for assessing training programs targeting health care professionals [24]. The Kirkpatrick model evaluates training programs based on 4 categories: the satisfaction of the participants with the training, improvement of their knowledge, improvement of their care practices, and improvement of patient health. The first category is very important in the evaluation process because, according to Kirkpatrick, learner appreciation is a key factor in motivating participants to learn from training and apply what they learned.

Therefore, based on our analysis of the Cochrane review of interventions for increasing the use of SDM by health care professionals, we first sought to determine whether the SDM training programs for health care professionals that used reflexivity strategies were more frequently classified as effective than the training programs that did not. Second, if such programs were more effective, we aimed to explore the strategy (peer-to-peer group learning or self-appraisal individual learning) that seemed to be more effective. Third, we aimed to examine whether training programs with an interprofessional orientation tended to be evaluated as effective. Finally, we examined whether the type of measurement used (OBOM or PROM) made it possible to classify the training programs in terms of effectiveness.

We performed a secondary analysis of a published 2018 Cochrane review that aimed to evaluate the effectiveness of interventions for increasing the use of SDM by health care professionals [7]. As there are no reporting guidelines for the secondary analyses of systematic reviews, we used the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 for all applicable items [25]. This study took place from February 2021 to April 2022.

The search strategy for the Cochrane review serving as the basis of the current secondary analysis was launched on June 15, 2017. Details of the Cochrane search strategy and data sources can be found in the published review [7]. The Cochrane review [7] included interventions classified according to the three target categories of the Effective Practice and Organization of Care (EPOC) taxonomy of interventions [26]: (1) interventions targeting patients (eg, patient-mediated interventions), (2) interventions targeting health care professionals (eg, distribution of printed educational material, educational meetings, audit and feedback, reminders, and educational outreach visits), and (3) interventions targeting both patients and health care professionals (eg, a patient-mediated intervention combined with an intervention targeting health care professionals).

In the published Cochrane review, the participants in the training programs could be any type of health care professional (eg, physicians, nurses, pharmacists, or social workers), including professionals in training (eg, resident physicians). Studies that recruited eligible health care professionals along with other types of participants (eg, patients and managers) were also included, as were training programs evaluating a multicomponent intervention (eg, SDM training for health care professionals with the use of patient decision aids). All types of training formats were incorporated into the review (eg, in class, group workshop, web-based training, and synchronous or asynchronous training). A training program was defined as a capacity-building activity conducted live for a group or a single individual, such as a web-based course or a traditional course (ie, a course integrated into an academic program), that used a recognized instructional method such as lectures, workshops, case studies, demonstrations, role plays, and small group discussions [10]. There were no restrictions on the comparison groups, which were all included.

The same primary outcomes of interest reported in the Cochrane review, namely SDM outcomes, were maintained in this secondary analysis, as were the outcome measurements used [7]. The Cochrane review grouped secondary outcomes into 2 categories: patient outcomes (eg, affective-cognitive outcomes, behavioral outcomes, and health outcomes) and process outcomes (eg, consultation length, costs, and equity). The 4 eligible study designs were randomized controlled trials, nonrandomized controlled trials, controlled before-and-after studies, and interrupted time series.

In this secondary analysis, we selected all interventions targeting health care professionals from the Cochrane review (43/87, 49%), excluding interventions that only targeted patients (44/87, 51%). Of the 43 that targeted health care professionals, we excluded interventions that were not a training program (12/87, 14%), for example, demonstrations on how to use a decision aid. Therefore, the total number of articles included in this study was 31.

One of the reviewers (NTD) examined all the content available concerning the interventions used in each of the 87 articles included in the Cochrane review and selected interventions that met the inclusion criteria [7]. Another reviewer (JL) reviewed both the selected and excluded programs to validate the rigor of the selection process. Finally, all the articles involving an eligible training program targeting health care professionals were selected (N=31) (refer to the eligibility criteria section).

Data extraction was performed by 2 pairs of independent reviewers (NTD and AM or NTD and VB). Differences between the 2 reviewers were resolved by consensus based on discussion and by referring to the definitions provided in the extraction grid (Multimedia Appendix 1). For the remaining conflicts, a third reviewer (SG-B or MCT) intervened to facilitate a consensus. The articles believed to include reflexivity were submitted to MCT for validation. Data extracted included (1) article and study characteristics—year, country and language of publication, and measurement of study design and type; (2) information on the training program—country, training language, context of care, type of health care professional trained, and training format (eg, unidirectional or interactive); and (3) the use or otherwise of reflexivity strategies for interactive training programs and, among those including reflexivity strategies, the type of strategy selected (peer-to-peer reflective group learning or self-appraisal learning).

We determined whether the included training programs were classified as effective in the Cochrane review [7]. The same outcomes of interest reported in the Cochrane review, that is, SDM outcomes, were maintained in this secondary analysis.

First, unidirectional and interactive training programs were compared to see what percentage of each was classified as effective by the Cochrane review. Second, among the interactive programs, those that used reflexivity strategies were compared with those that did not to see which were more frequently effective. Third, reflexivity strategies (peer-to-peer group learning and self-appraisal individual learning) were compared. After these 3 steps, we carried out an additional analysis to see whether there were elements that could explain why a training program was effective or otherwise, apart from those cited earlier. For example, we compared the proportion of training programs based on interprofessional orientation that were effective with the proportion of programs using reflexivity that were effective. In addition, to see how SDM training programs can be better evaluated, we classified training programs according to the type of measurement they incorporated and whether they were classified as effective.

The risk of bias assessment was reported in the initial Cochrane review, which used criteria for EPOC reviews [73] and the Cochrane Handbook for Systematic Reviews of Interventions [74] for interrupted time series designs.

For our variable of interest (the primary outcome), the certainty of evidence was assessed in the Cochrane review according to the GRADE (Grading of Recommendations Assessment, Development and Evaluation) guidelines and methods described in Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions [75]. The method proposed in EPOC worksheets was used to determine which secondary outcomes should be assessed [76].

This secondary analysis of a 2018 Cochrane review on interventions for increasing the use of SDM by health care professionals aimed to identify training programs that included reflexivity and to ascertain how effective they were. Our study is the first to evaluate the effectiveness of SDM training programs that include reflexivity strategies. Among the 31 SDM training programs for health professionals, 23% (n=7) included reflexivity and 77% (n=24) did not. More of those that included reflexivity were deemed effective as a percentage of the whole (3/7, 43%) than those that did not (8/24, 33%). Among the interactive training programs, there was little difference in effectiveness between those that used reflexivity strategies (3/7, 43%) and those that did not (7/17, 41%). However, when comparing interactive training programs with unidirectional ones, there were a great many more programs deemed effective among the former (10/24, 42%) than among the latter (1/7, 14%).. Among the training programs that included reflexivity (n=7), most programs using a peer-to-peer group learning strategy were found to be effective (3/5, 60%), whereas those using a self-appraisal individual learning strategy were not (0/2, 0%). The training programs with an interprofessional orientation were more frequently classified as effective (3/5, 60%) than those without (2/5, 40%). Finally, the percentage of effective training programs in studies using OBOMs to assess training was higher than the percentage in studies using PROMs. These findings led to the observations that follow.

Our results confirm the findings of Leyland et al [77] and Chaffey et al [12] that in general, programs that include reflexivity have more positive effects than those that do not [12,77]. Training using reflexivity strategies has been shown to increase medical students’ ability to integrate alternative sources of knowledge and critically reflect on their own practices [12]. According to Chaffey et al [12], it is difficult to assess reflexivity alone, but when applied in an intervention, it yields more positive results. Although only a few SDM training programs include reflexivity strategies to date, these results suggest that reflexivity could be a core component of effective training and, as such, may be understood as an effective implementation strategy for change. However, further studies with a larger sample are needed to confirm this hypothesis. For Kolb [78], learning is a complex process driven partly by individuals’ ability to find a sense for themselves in operationalizing change in their day-to-day practices. As our results suggest, training that uses reflexivity strategies could provide health care professionals with that personal sense of the purpose of SDM and the motivation needed to implement SDM in their practices. However, future research could more precisely evaluate the impact of reflexivity strategies on SDM uptake by health care professionals if more programs were explicit in their use or promotion of reflexivity. The training programs included in this review were not designed to use or promote reflexivity. An analysis that compared SDM training programs that were explicitly based on reflexivity with those that were not would be more useful and appropriate. Nevertheless, our results suggest that SDM training for health care professionals based on reflexivity strategies is effective in motivating trainees to adopt SDM in a manner that facilitates positive patient experiences in health care systems.

Our findings also showed that more training programs using peer-to-peer group learning were classified as effective than those using self-appraisal individual learning. These findings suggest that a self-appraisal strategy can be more effective than training focused on an individual if it is part of a group learning process. Interaction among learners seems to be a powerful strategy for encouraging reflection, even self-reflection. Research could further compare the types of reflexivity learning strategies that are most effective in SDM training for health care professionals.

Our findings indicated that training programs with an interprofessional orientation have proven to be more effective than those without this orientation. All programs that used interprofessional orientation and reflexivity strategies were also deemed effective. In addition, these programs used a peer-to-peer group learning strategy in which participants were encouraged to learn from each other’s reflections by sharing their points of view and experiences [28]. One of the main goals of interprofessional SDM is to encourage the recognition of other professionals’ values and competencies [21,79], in other words, to recognize other members of the care team as peers [22]. Leyland et al [77] also considered that reflexivity is needed for health care professionals to recognize each other’s professional skills. This suggests that the peer-to-peer group learning strategy is especially appropriate for implementing an interprofessional approach. In addition, Tremblay et al [27] defined the goal of “formative reflexivity” (ie, reflexive practices for learning) as developing new visions of professional experience and working in collaboration. Thus, blending these 2 goals (interprofessional orientation and reflexivity) is a promising avenue for fostering the uptake of interprofessional SDM in health practice. Furthermore, encouraging reflection and questioning of one’s practice can help to correct the paternalistic way in which SDM still seems to be undertaken [19,77].

The Cochrane review used only OBOMs and PROMs (or both) to evaluate intervention effectiveness, eschewing programs that used HCPROMs. In the review, most programs classified as effective were assessed using OBOMs (8/11, 73%) rather than PROMs (3/11, 27%). OBOMs are considered a more rigorous assessment option than self-reported measures because they are more independent [23]. Based on this justification, one can say that the apparent objectivity of OBOM assessment yields better results than the potential subjectivity arising when patients report an outcome. Furthermore, our findings showed that 2 (67%) of the 3 training programs using reflexivity and classified as effective were assessed with OBOMs, as opposed to only 1 (33%) with PROMs. However, neither PROMs nor OBOMs respect the first Kirkpatrick model criterion, which is the satisfaction of the participants with the training. According to Kirkpatrick, when participants like a training program, they may be willing to adopt what they have learned from it. Based on the importance of this first level of evaluation, we suggest that the perspective of the training beneficiaries (health care professionals) should be considered in evaluations, as is the case with HCPROMs. This does not exclude the use of PROM or OBOM measures to assess Kirkpatrick’s other 3 levels of improvement: knowledge improvement, practice improvement, and health-related outcome improvement (the levels considered in the Cochrane review). Furthermore, a reflexivity-based approach focuses not only on outcomes but also on the process or experience of health care, which may be as important to patients as outcomes [80]. If trainees reflect on their own practice and question their own values, they will better consider patients’ values and experiences in the health care-seeking process. Therefore, another relevant measure of effectiveness would be patient-reported experience measures (PREMs), which help improve patient care [81]. In a 2019 systematic review, Müller et al [9] analyzed the methods used by 41 studies that assessed SDM training programs for health care professionals and concluded that the diversity of assessment methods limits the ability to compare training program effectiveness and is a barrier to conclusive evidence. Therefore, it now seems important to develop a harmonized SDM training assessment measure that includes all the 4 perspectives to enable better comparison.

In this review, some interventions involved multiple components, for example, they included a workshop, a web-based tutorial, and a decision aid tool or a workshop with audit and feedback. In these cases, it was difficult to evaluate the effects of the components separately. In addition, some of the training programs included patients. If the training includes patients, it can be difficult to know whether this inclusion has an additional impact on effectiveness. Another example is the diversity of the comparators. The articles included in the review used different types of comparators (usual care or another differing intervention). Finally, even if the primary outcome (SDM) was our focal interest, it may be defined differently from one study to another. For example, while some focus their analysis on the uptake of SDM as a whole, others may analyze only one component (eg, decision conflict or decision regret) or separate them. In some articles, analysis used ≥2 scales, which made our judgment difficult, especially if effectiveness-related results were different. To classify these types of articles in terms of effectiveness, we referred to the conclusions of the authors. Based on all these observations, we suggest using a core set of assessment methods and validated outcomes for all learning levels in SDM training programs to identify the most effective strategies and better compare them. Future research could explore methods to specifically assess reflexivity strategies included in SDM training programs targeting health care professionals for their inclusion in this core model. Our reflexivity grid, when validated and published, along with the results from this analysis, can be considered as the first step in assessing SDM training that uses reflexivity strategies, but further work is needed to guide training approaches in this field.

This study has a few limitations. First, the systematic review included evidence only up to June 2017, and another update has not been performed since then. The Cochrane review has been updated twice, with each update including all the studies in the earlier versions; yet, its conclusions regarding effectiveness have varied little since the first review in 2010. At its first publication, it included only 5 RCTs, but in 2018, with 84 RCTs, it still concluded that “a great variety of activities exist to increase shared decision-making by health care professionals, but we cannot be confident about which of these activities work best because the certainty (or the confidence) of the evidence has been assessed as very low.”

Second, seeing that the training programs included in the systematic review were not explicit about promoting reflexivity, the selection of those that included what could be accurately described as reflexivity strategies was not an easy task. Nevertheless, using our grid based on a preliminary rapid review, we identified 7 programs. This was a small number for further analysis, although not surprising, as reflexivity is a new concept in the context of health professional education about SDM. It is possible that we missed a few programs that included reflexivity, as we relied solely upon published data and did not contact the authors for additional information. To minimize this risk, we analyzed all additional materials cited in the articles that were linked with the training programs and considered the most recent publications to discuss our results. Finally, the grid used in the analysis was developed for the purpose of this study, and although it is an important contribution to the field of implementation, its validation will require further use and assessment.

Our results suggest that SDM training programs for health care professionals using reflexivity strategies could increase SDM implementation. Our study is the first to evaluate the effectiveness of SDM training that includes reflexivity and raises several questions: (1) Are peer-to-peer learning strategies more effective than self-appraisal strategies? (2) How can reflexivity and interprofessional orientation strategies best complement each other? (3) Are OBOMs and PROMs the only appropriate means of evaluating SDM training programs? The grid developed for identifying reflexivity strategies in training programs, including reflexivity-related approaches and concepts, will be a useful guide for developing reflexivity training and is to be validated in future studies.