Building a data-driven evaluation framework of physician wellness initiatives: a rapid review Authors: Izzah Khairi, Sridevi Kundurthi, Brian Lo, Tania Tajirian, Terri Rodak, Sanjeev Sockalingam, Treena Wilkie Published: 2026 ABSTRACT AbstractObjectivesWhile there exist many individual and organisation-level initiatives aimed at reducing physician burnout and promoting wellness, there are no comprehensive frameworks or guidelines for evaluating initiatives targeted at physicians. To address this gap, we conducted a rapid review to understand the current state of initiative evaluation in this field and develop an evaluation framework for initiatives aimed at physician burnout and wellness.DesignA rapid review based on the Cochrane rapid review methods guidance.Data sourcesMEDLINE, Embase and PsycInfo were searched from database inception to 2 May 2024 for concepts related to physicians, wellness initiatives and burnout, and organisational efforts.Eligibility criteriaWe included studies with initiatives aimed at physician burnout, wellness or experience; targeted physicians, residents, fellows and/or physician faculty; were evaluated in some format; and took place within healthcare settings.Data extraction and synthesisTwo independent reviewers extracted data according to a standard template. For each study, we noted information related to the type of wellness initiative, evaluation approach, components measured in evaluations (‘evaluation indicators’) and gaps in evaluation. The same reviewers analysed the data quantitatively and thematically. Findings were reported according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews.ResultsThe database search yielded a total of 3786 references, of which 105 were included based on predefined inclusion and exclusion criteria. Various types of wellness initiatives (eg, mindfulness, peer support programmes, communities of practice) were delivered multimodally, or via curriculums and workshops, among other formats. Common evaluation approaches included surveys (95%, n=100/105) and interviews or focus groups (17%, n=18/105). Evaluation indicators spanned five categories: impacts (94.3%, n=99/105), participants’ reactions (65.7%, n=69/105), perspective and behaviour changes (46.7%, n=49/105), implementation (45.7%, n=48/105) and continuous improvement (6.7%, n=7/105). Evaluation gaps included limited responses from participants, failure to capture certain indicators (eg, physician turnover) and limited longitudinal measures.ConclusionsBased on the findings and existing models, we developed a physician-focused evaluation framework, constituting two domains: implementation and impacts. This framework can enable organisations to better understand, assess and improve initiatives aimed at physician well-being, which can have positive impacts on patient care and the healthcare system.Keywords: Physicians, Burnout, Implementation Science, Quality Improvement Strengths and limitations of this study: Using a comprehensive search strategy across three major databases (MEDLINE, Embase, PsycInfo) captured a wide range of literature on the current state of evaluating physician wellness initiatives. Using a comprehensive search strategy across three major databases (MEDLINE, Embase, PsycInfo) captured a wide range of literature on the current state of evaluating physician wellness initiatives. However, by focusing our search specifically on physician wellness initiatives, relevant evaluation metrics and frameworks from related fields may have been missed. However, by focusing our search specifically on physician wellness initiatives, relevant evaluation metrics and frameworks from related fields may have been missed. Introduction Burnout is defined as a syndrome resulting from chronic work-related stress involving three dimensions: emotional exhaustion, depersonalisation and reduced personal accomplishment.1 Studies from North America demonstrate significantly higher rates of burnout among physicians compared with other healthcare professions, highlighting a particular concern for this population.2–5 In 2017, the prevalence of burnout among US physicians, for example, was 39.8% compared with 28.1% of working adults.5 This high prevalence of burnout has persisted over time, with 46% of Canadian physicians experiencing significant burnout in 2003,6 and 53% of Canadian physicians reporting symptoms of burnout in 2021.7 This long-standing issue has implications for patients, the healthcare system and physicians themselves, as burnout is associated with lower career engagement,8 higher occupational turnover,8–10 worse quality of patient care8–10 and increased healthcare system costs.9 11 12 Given these adverse effects, several individual-level and organisation-wide initiatives have been implemented with the aim of addressing physician burnout.13–15 Accompanying these initiatives are efforts evaluating their effectiveness. Evaluation methods geared towards determining the effectiveness of these interventions have included surveying physicians at baseline and post-initiative to assess for any changes due to the initiative16 17 or using interviews to supplement survey data.18 Evaluation indicators, such as fatigue or job satisfaction, also varied widely, reflecting differences in the objectives of wellness initiatives and the heterogeneity in approaches to evaluate said initiatives.9 Frameworks have also been used to guide evaluations, with the most well-known of these being logic models and the RE-AIM (Reach, Effectiveness, Adoption, Implementation and Maintenance) framework.19 20 To date, however, there is an absence of frameworks specifically tailored to evaluating wellness initiatives for the physician population. Given the complex interplay of intrinsic, external and organisational factors shaping physicians’ experiences and burnout,10 there is a need for evaluation models that incorporate this complexity. As well, evaluating physician wellness initiatives through a physician-centric lens is essential to ensure these efforts are relevant and responsive to physicians’ lived experiences. With the growing number of physician wellness initiatives, the impacts of these initiatives have been widely documented in the literature; however, there remains limited guidance on how these initiatives should be systematically evaluated. Thus, in this rapid review, we aimed to identify key components of approaches currently used to evaluate wellness initiatives and subsequently leverage these insights to create an evaluation framework that can aid organisations in assessing initiatives supporting physician well-being. Methods A rapid review was conducted in accordance with the Cochrane rapid review methods guidance,21 and results are reported according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidance (see online supplemental file 1).22 bmjopen-16-4-s001.pdf Information sources and search strategy Searches were conducted in the following three databases on 2 May 2024: MEDical Literature Analysis and Retrieval System OnLINE (MEDLINE), Embase, American Psychological Association (APA) PsycInfo (see online supplemental file 2 for complete MEDLINE search strategy). A health sciences librarian (TR) developed the search strategy in MEDLINE in collaboration with study authors (TW, BL), then translated the search for use in other databases. All strategies used database-specific subject headings, keywords in natural language and advanced search operators to capture three concepts: (1) physicians and medical trainees (eg, ‘physician,’ medical resident, specific medical specialists), (2) wellness initiatives or burnout (eg, ‘wellbeing’, ‘burnout’, ‘compassion fatigue’) and (3) organisational efforts (eg, ‘program evaluation’, ‘quality improvement’, ‘implementation’, ‘framework’). No study type or publication date limits were applied. bmjopen-16-4-s002.pdf Inclusion and exclusion criteria Studies were included if they described initiatives with the objective of addressing burnout, wellness or experience, and evaluated this initiative (eg, through surveys, interviews, etcetera); described initiative(s) that targeted physicians, residents, fellows and/or physician faculty; were conducted in healthcare settings, such as hospitals or community clinics; and were written in English. Studies were excluded if they described initiatives that were not focused on addressing physician burnout, wellness or overall experience; only targeted medical students; were solely focused on addressing electronic health record (EHR)-related burnout; were based in non-healthcare settings; did not reference evaluation; or were classified as reviews, conference papers, commentaries, letters, editorials, protocols and dissertations. Screening process All studies resulting from database searches (n=3786) were uploaded to Covidence. Two authors (BL, IK) conducted pilot screens by reviewing 100 articles and resolving discrepancies to refine inclusion and exclusion criteria. Subsequently, one reviewer (IK) screened the titles and abstracts of all studies based on this revised criteria, followed by a full-text review of the remaining studies. When the reviewer had difficulty determining whether a study should be included, this was resolved through discussion with study team members (SK, BL, TW). Data extraction and synthesis A data extraction template was collaboratively developed (IK, SK, BL, TW) and included study characteristics such as publication year, setting, country, population and study design. To address the research question, the following information was collected: initiative type(s), evaluation data collection approaches, quantitative and qualitative evaluation indicators, evaluation gaps and learnings and presence of an evaluation framework (see online supplemental file 3 for complete extraction template). Study team members (IK, SK) analysed data quantitatively (to compute descriptive statistics) and thematically (to categorise evaluation indicators, gaps and learnings). bmjopen-16-4-s003.pdf Results Study characteristics Applying our search strategy to the databases yielded a total of 3786 references. Of these, 178 studies underwent full-text review, and 73 articles were excluded. A total of 105 articles were deemed eligible (see figure 1). Table 1 provides an overview of included studies’ characteristics. Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram. Table 1Study characteristicsCharacteristicN (%)ReferencesYear 2024 to 202062 (59.0%) 23 25–27 32 33 35–37 39–88 125 126 130 2019 to 201531 (29.5%) 24 28 34 89–114 123 124 2014 to 20105 (4.8%) 29 38 115–117 2009 to 20056 (5.7%) 30 118–122 2004 to 20000 (0.0%)N/A Before 20001 (1.0%) 31 Study location USA62 (59.0%) 23 24 28–31 33–36 38 39 45 47 49–58 60 61 63–67 70 72 73 76–78 80–83 85 86 91 92 95 96 98 100 105 107 112–114 117 118 121–123 125 126 130 Australia13 (12.4%) 25 32 69 71 84 87 90 101 104 110 111 119 124 Canada8 (7.6%) 27 62 75 79 89 93 106 108 UK8 (7.6%) 37 41 44 46 74 99 103 109 The Netherlands3 (2.9%) 40 59 120 Germany2 (1.9%) 43 115 Japan2 (1.9%) 68 94 Other countries (n=1 each)7 (6.7%)Qatar; China;27 Tunisia; Saudi Arabia; Spain; Romania; Egypt26 48 88 97 102 116Population Trainees (residents and fellows)52 (49.5%) 23 26 28 30–32 34 36 37 42 43 52–55 57 59–61 63 64 66–69 71 73 82 83 87–91 93–95 100 103–106 109–111 114 117 118 121–124 Post-training physicians24 (22.9%) 24 25 27 29 35 39 40 45 49 58 62 79 81 85 86 92 97 99 102 108 115 116 119 130 Mixed (physicians/trainees and others)29 (27.6%) 33 38 41 44 46–48 50 51 56 65 70 72 74–78 80 84 96 98 101 107 112 113 120 125 126 Setting Academic/university hospital(s)76 (72.4%) 23–25 27–32 34–36 38 39 42–44 48 49 51–59 61–67 69–73 75 76 78–80 82 84 87–91 93 95 96 98 100 101 103–108 112 113 116–119 121–125 130 Multi-site (eg, community- and hospital-based, across entire geographic region)14 (13.3%) 26 33 40 45 46 50 60 74 81 97 99 109 114 115 Non-academic hospital(s)7 (6.7%) 41 68 83 92 94 102 120 Professional organisation*6 (5.7%) 37 85 86 110 111 126 Clinic(s)2 (1.9%) 47 77 Study design Quasi-experimental52 (49.5%) 23 24 26 31 34–36 38–40 43 44 47–51 53 54 56 60 63 64 66 69 71 72 74 76 78 84 88 91 92 95–99 102–104 107–110 112 114 116 120 122 126 Randomised controlled trials22 (21.0%) 27 29 30 33 37 42 45 46 55 58 61 65 70 77 85 87 89 90 106 111 119 124 Cross-sectional12 (11.4%) 25 28 57 68 73 79 94 100 115 121 123 125 Mixed methods10 (9.5%) 41 52 62 75 81 83 86 101 105 113 Cohort8 (7.6%) 32 59 67 80 82 93 117 118 Article1 (1.0%) 130 *Professional organisations refer to colleges, specialties, fellowships, etc that coordinate programmes, training or initiatives for physicians practising in healthcare settings. Study characteristics *Professional organisations refer to colleges, specialties, fellowships, etc that coordinate programmes, training or initiatives for physicians practising in healthcare settings. Wellness initiatives For each of the included studies, we classified the type of wellness initiative delivered (see table 2). The number of individuals participating in the initiatives ranged from 623 to 1537.24 Nearly one-fifth (n=19) of the initiatives delivered included more than one initiative, or were ‘multimodal.’ For example, the hospitals in Petrie and colleagues’25 study incorporated individual-level strategies, such as mental health training and mentoring programmes, in addition to organisational-level strategies, such as the formation of a wellness committee and employing more staff to reduce overtime. 18 initiatives (17.1%) were educational programmes and were held on a variety of wellness-related topics, such as stress,26–31 resilience32 33 or leadership.34 16 initiatives (15.2%) focused on mindfulness (including yoga, self-reflection and mindfulness-based stress reduction programming), while another 13 (12.4%) were related to work hours and shift scheduling. Mentorship encompassed mentoring and coaching programmes and constituted 10 initiatives (9.5%). Communities of practice, such as Balint groups and Schwartz rounds, comprised 9 initiatives (8.6%), as did peer support programmes. A further 11 (10.5%) initiatives (labelled ‘other’) did not fit into any of the above categories, including nature therapy,35 36 text messaging for sharing resources37 and incentivised exercise programmes,38 among others. Table 2Types of physician wellness initiativesWellness initiative typeN (%)DescriptionReferencesMultimodal initiatives19 (18.1%)Wellness programmes that include more than one initiative 25 44 50 52–54 59 62 66 69 71 72 76 78 79 87 95 101 113 Educational programmes18 (17.1%)Initiatives teaching about wellness or wellness-related topics, primarily through workshops and curriculums 24 26–34 39 47 60 77 111 116 119 126 Mindfulness initiatives16 (15.2%)Initiatives bringing to attention an individual’s awareness of themselves (eg, thoughts, emotions, senses) and the present moment 40 43 46 49 51 55 56 61 92 97 99 103 104 108 112 Work schedules13 (12.4%)Initiatives related to the number of work hours or scheduling of shifts 68 82 88 89 93 94 106 115 117 118 121–123 Mentorship10 (9.5%)Initiatives where participants are paired with a mentor or coach 45 63 70 74 85 91 98 100 110 124 Communities of practice9 (8.6%)Groups composed of individuals from similar backgrounds or with a common purpose. They may be multidisciplinary in nature 23 42 58 65 90 102 105 109 120 ‘Peer support’ initiatives9 (8.6%)Initiatives where individuals in distress are connected to peers 41 57 73 80 81 84 86 107 125 Other11 (10.5%)All other initiative types 35–38 48 64 75 83 96 114 130 Types of physician wellness initiatives Evaluation For each included study, we noted methods of data collection, indicators used to measure the evaluation of an initiative, frameworks used to guide the evaluation, and evaluation gaps and learnings. Data collection approaches Studies collected data to evaluate their initiatives in a variety of ways. Data were collected using surveys (95.2%, n=100/105)23–122; interviews or focus groups (17.1%, n=18/105)27 30 50 52 60 62 65 69–71 75 83 87 94 105 109 123 124; documents (12.4%, n=13/105)41 43 50 51 61 85 90 96 105 113 117 118 125 such as administrative data on the initiative costs43 61 85 96 105 113 and facilitator notes41 90; and other approaches (11.4%, n=12/105)43 46 61 67 73 82 89 101 109 112 119 125 such as participant observation61 109 and remote monitoring for sleep.67 82 The majority of studies using surveys collected data pre-initiative and post-initiative (67.6%, n=71/105).24–40 43 44 46–56 59–61 64 66 67 69 71 72 76–78 80 84 87 88 92–97 99–101 103–108 111–119 121 Common post-initiative time collection points for surveys were immediately after the end of the initiative (22.5%, n=16/71),23 26 32 35 36 38 45 58 61 69 92 97 106 108 111 120 at 3 months (23.9%, n=17/71),24 27 30 32 34 37 39 45 56 65 85 91 96 99 107 111 120 6 months (18.3%, n=13/71),27 33 42 45 54 58 69 73 86 111 116 119 120 and 12 months (21.1%, n=15/71).33 39 50 53 54 77 78 80 91 98 110 111 119 121 126 18 studies conducted interviews and focus groups, which were done either only post-initiative (88.9%, n=16/18),27 30 50 52 60 62 69–71 75 83 87 94 105 123 124 except for a few that also conducted interviews pre-initiative (11.1%, n=2/18).43 96 In total, over 100 scales were used across 85 studies (80.9%, n=85/105).23–40 42–58 60–65 68–79 82 83 85 87 89–92 94–100 102 104 106–108 111 112 114–122 126 See table 3 for scales used to assess burnout and well-being. Table 3Burnout and well-being scalesOutcome measuredScaleN (%)ReferencesBurnoutMaslach Burnout Inventory (MBI) and its variants (eg, aMBI, MBI-HSS)56 (53.3%) 24 26 28 30 31 33–35 38 40 44–48 51 53–58 61–63 65 68 70 73 78 79 82 85 87 89–92 94 95 98–100 102 108 112 114–122 126 Professional Quality of Life6 (5.7%) 32 44 87 96 107 111 Stanford Professional Fulfillment Index5 (4.8%) 23 49 61 77 85 Copenhagen Burnout Inventory5 (4.8%) 37 43 69 71 104 Other burnout scales (n=1 each)3 (1.0%)Physician Work Life Survey; Burnout Clinical Subtype Questionnaire; Cognitive Hardiness Scale83 97 122Well-beingMayo Clinic Well-Being Index4 (3.8%) 4 23 39 50 WHO-5 Well-Being Index3 (2.9%) 69 71 104 Short Warwick-Edinburgh Mental Wellbeing Scale3 (2.9%) 37 46 74 Resident Well-Being Index2 (1.9%) 61 64 Other well-being scales (n=1 each)8 (7.6%)Mental Health Continuum; Life Satisfaction Questionnaire; Resident Wellness Scale; Psychological General Wellbeing Index; Positive Emotion, Engagement, Relationships, Meaning and Accomplishments; Stanford Physician Wellness Survey; Dupuy Psychological General Well-Being Scale; Brief Resident Wellness Profile40 52 54 63 77 95 106 Burnout and well-being scales Evaluation indicators Authors used various components and metrics to evaluate their initiatives, referred to as evaluation indicators (see table 4). We categorised evaluation indicators into five groups: (i) impacts (94.3%, n=99/105); (ii) participants’ reactions (65.7%, n=69/105); (iii) perspective and behaviour changes (46.7%, n=49/105); (iv) implementation (45.7%, n=48/105); and (v) continuous improvement (6.7%, n=7/105). Within each group, we further categorised indicators into primary categories. See table 3 for definitions and categories of each grouping. Table 4Groupings of evaluation indicatorsGroupingN (%)DefinitionPrimary categoriesReferencesImpacts99 (94.3%)Indicators related to broader outcomes of interest to stakeholders, at both the individual-level and organisational-levelPhysician’s health, patient care, work-related, burnout/wellness, diagnoses, personal life, organisation-level, unintended consequences 23–35 37–83 85–104 106–109 111–124 126 Participants’ reactions69 (65.7%)Indicators eliciting feedback from the participants’ of an initiative, and other opinionsParticipants’ awareness of initiative, participants’ feedback, facilitators to success, challenges/barriers to initiative’s success, continued or future participation, improvements, factors affecting burnout, well-being or stress 24 26 30–32 35 36 39–44 46 47 50 52 54 55 57 59–62 65–67 70–73 75 78–81 83–86 88 91 93–96 98–101 104–112 114 116–123 125 Perspective and behaviour changes49 (46.7%)Indicators assessing changes in participants’ own learnings or actionsKnowledge changes, attitude changes, skill changes, behaviour changes 24–29 33–36 39–41 43 45–47 49 51 55 60 61 63 66 70 77 81 85 87 89 91 92 95 97 98 100 101 103 104 106 108 109 111 112 114 120 123 126 Implementation48 (45.7%)Indicators measuring how the initiative is formatted, delivered and receivedStage of implementation, formal and informal support, delivery, costs, participation metrics, evaluation time, other 25 27 38–41 43 44 47 49 51 56–59 61 63 64 70 73 75 76 78–81 84–86 91 92 96 98 100–105 107 109 111 113 120 122–125 Continuous improvement7 (6.7%)Indicators assessing an initiative’s ability to be delivered as intended, adapt to changing conditions, and maintain or expand its operations over timeFidelity, flexibility, sustainability 43 45 46 57 61 68 81 Groupings of evaluation indicators Evaluation models Nine studies (8.5%) used established evaluation models to assess their initiatives. Three studies79 81 103 used the PRISM (Practical, Robust Implementation and Sustainability Model) and/or RE-AIM framework20 127; two studies75 101 used logic models19; a further two studies43 61 used Bowen and colleagues’128 areas of focus for feasibility studies; one study26 used the Kirkpatrick Model129; and another130 developed its own Well-Being Evaluation Schema (WES). Evaluation gaps and learnings We also recorded the study authors’ perceived strengths and limitations of their evaluations. 38 of 105 (36.2%) studies reported that a gap in their evaluation was the minimal number of responses from participants who had accessed or engaged in the initiative,24 25 27–29 33 34 38 39 41 44 47–50 52 53 59 63 66 69 71 73 80 81 85 92 98 99 103 107 108 110 111 114 117 120 122 with the lowest reported rate being 9.8% from 4 out of 41 respondents.94 In addition, some studies had high pre-initiative assessment response rates, but lower response rates at follow-up; for example, a 100% response rate of 664/664 at baseline that declined to 46% with 304/664 at 6 months.120 Some authors attributed the low response rates to busyness, lack of financial incentives and cluttered email inboxes for online surveys, as well as the voluntary nature of the evaluation.43 65 80 81 84 92 Limited response rates could present an evaluation problem due to potentially missing other perspectives from the general participant sample.33 122 Another limitation noted by study authors was not using a wider range of approaches and tools for data collection. Two studies noted a lack of qualitative data in their evaluations, as they only used quantitative methods,42 91 whereas another study stated they could have used quantitative scales alongside qualitative approaches.84 Of the reviewed studies, Ward and colleagues69 advocated for evaluating programmes with a mixed-methods approach, where quantitative outcomes could be complemented by including qualitative data to provide insight into the ‘impact, acceptability and utility’ of initiatives. As the most commonly used evaluation approach, surveys provided certain strengths and limitations. One study highlighted their convenience for assessing changes over time in a large sample,120 while another study noted the potential of survey fatigue.64 Three other studies indicated that holding interviews or focus groups in addition to surveys may have strengthened their evaluation or provided more in-depth data on the wellness initiatives.66 79 110 Studies using surveys often included validated scales to measure indicators; in fact, eight studies stated that a strength of their evaluation was in their usage of validated scales, which may have helped reduce subjectivity.27 44 48 53 57 64 117 118 Conversely, other authors used non-validated scales25 26 96 103 107 or shortened instruments47 112 that may not have been able to fully assess outcomes. Study authors also identified evaluation indicators that they failed to measure. These included demographic data (eg, gender,42 63 career stage76 etc), implementation indicators (eg, fidelity,45 106 125 satisfaction,79 97 etc), post-implementation indicators (eg, physician turnover,42 45 62 74 81 120 burnout,29 81 105 etc) and external factors (eg, pandemic,70 80 82 social support,28 102 etc). Furthermore, the majority of studies measured outcomes with self-report data, which Fendel and colleagues43 noted as contributing to their high number of responses for their evaluation; however, others identified that their evaluation was limited by the absence of objective data.45 55 66 88 93 120 123 Only four studies conducted a needs assessment prior to the initiative to ensure that individuals’ needs were identified and addressed with the initiative, which the same studies also later evaluated.43 62 75 101 For some studies that collected data post-initiative, authors identified that a limitation in their evaluations was the study’s short timeframe, and that they could have benefitted from conducting their evaluation over a longer period, such as beyond a year, to assess for long-term changes.24 25 29 32 36 38–40 43 48 51 53 65 72 74 79 81 85 95 97 103 104 109 124 Alternatively, a few authors measured burnout longitudinally pre-initiative and post-initiative, but regretted not assessing burnout for short-term effects,37 77 112 as using only annual assessments may be too ‘blunt’ to detect immediate changes resulting from the initiative.77 Discussion In the present review, we aimed to find studies evaluating burnout and wellness initiatives for physicians, with the goal to categorise evaluation approaches and indicators, and use these insights to produce an evaluation framework. We found a wide range of approaches, indicators and models that were used to evaluate physician-specific wellness initiatives. Our findings highlight similarities in evaluations across initiatives. The most common data collection tool was surveys, which were used by 95% of the studies. Their popularity may imply that surveys are most feasible for this population; however, some authors noted that a more holistic evaluation would integrate surveys with other forms of data collection, such as interviews for probing, or documentation review for objective indicators. Additionally, a challenge with surveys is with participants’ concerns of confidentiality, even when individuals are told that responses are anonymised, which may lead to potential for social desirability bias.131 These concerns may be further exacerbated by issues related to trust and organisational justice. While there were many different evaluation indicators in the included studies, burnout was measured by over 70% of studies and most commonly measured using the Maslach Burnout Inventory (MBI), including its variations, including the abbreviated MBI (aMBI) and MBI-Human Services Survey (MBI-HSS). Despite this indicator’s widespread usage, Guille & Sen132 note that the definition of burnout itself lacks consensus, resulting in differences across studies regarding what constitutes a physician meeting the criteria for burnout, which may lead to inconsistency in evaluation. Nearly one-fifth of the studies included multimodal initiatives, highlighting a growing trend towards addressing physician wellness through multiple strategies. As new opportunities for addressing physician wellness continually emerge, organisations could adopt a phased approach and add initiatives over time. While evaluating such initiatives, it may be difficult to isolate the impact of a single initiative from the potential synergistic effect of multiple initiatives. To disentangle this complexity, evaluations could include creating a programme theory of all wellness initiatives to understand how each individual initiative leads to specific outcomes.133 Of interest is that a few studies referenced the importance of incorporating non-participants in evaluation45 71 103 130; for example, to identify missing needs that are not currently being addressed by the programme.101 Incorporating feedback from individuals who want to participate but face barriers to do so, or individuals who may not feel the need to access the programmes but view them as beneficial for the community, could provide a more holistic view of the programme on physician experience at the personal and organisation levels. Furthermore, comparing changes in outcomes between participants and non-participants could provide more evidence for the initiative’s ability to affect change in physician wellness.133 Consolidating the evaluation gaps and learnings identified earlier, recommendations for evaluating physician wellness initiatives were consolidated. Using a mixed methods approach for collecting data can provide a more holistic evaluation; for example, supplementing surveys with interviews for more comprehensive feedback. Common measures missed as noted by authors of the included studies covered implementation indicators, such as fidelity45 106 125; post-implementation indicators, such as physician turnout42 45 62 74 81 120; and external factors, such as the extent of social support28 102; and demographic information, such as gender.42 63 Collecting this data can aid researchers to generate more comprehensive evaluations and tailor initiatives to the unique needs of different groups.46 61 76 The majority of evaluations collected data pre-initiative and post-initiative, which could help to ascertain whether initiatives meet participants’ needs, whether changes occurred as a result of the initiative, as well as long-term impacts of the initiative.133 There is some evidence supporting the usefulness of evaluating initiatives before implementation, including at the programme development stage,134 such as by consulting stakeholders prior to implementation, or creating a logic model prior to delivering the initiative to help providers anticipate how a programme’s activities will lead to desired goals.133 Lastly, low response rate can affect the generalisability of the evaluation. In the literature, suggestions for improving response rate among physicians include monetary incentives,135 136 personalised correspondences136 and shorter survey length,137 although these have received mixed results.138–140 The majority of studies in the review evaluated indicators under the impacts grouping, suggesting that assessing the outcomes of an initiative, including burnout, wellness and job satisfaction, is common indicators in the literature for physician wellness initiatives. This finding, however, is likely an artefact of our search strategy rather than reflective of a broader trend in the literature, since burnout and wellness were a focus for the studies we identified; thus, it follows that many authors ensured that they captured indicators measuring these outcomes of interest. Approximately half of the studies had indicators in the groupings of participants’ reaction, perspective and behaviour changes and implementation, which indicates an interest in evaluating how participants received the initiative, changes to participants’ own views and actions resulting from the initiative, as well as how the initiative was implemented. Only seven studies used indicators to measure continuous improvement, comprising indicators related to fidelity, flexibility or sustainability. This finding reflects a lesser focus on evaluating initiatives long-term and that the important indicators in this grouping may be missed in evaluations of physician wellness initiatives. A data-driven, physician-centric framework Of the included studies, nine used one of the following models to evaluate wellness initiatives for physicians: PRISM/RE-AIM, the Kirkpatrick Model, Bowen and colleagues’ feasibility areas, the WES, and a logic model. These models are not tailored for evaluating initiatives addressing physician burnout or well-being and were either too narrow or broad in scope for their application to the field of physician wellness. For example, the Kirkpatrick Model is exclusively intended for training programmes and may not be applicable to other forms of wellness initiatives. Alternatively, the WES, while targeted towards initiatives for physicians, was designed based on existing frameworks that focus on training and education; therefore, its scope may be less suited in assessing broader organisational or non-training initiatives. More broadly, these models do not account for the evolving needs of physicians over time, including changes across career trajectories and shifts in organisational culture. Given the limited scope of existing frameworks and lack of physician-specific considerations, we created our own framework for evaluating physician wellness initiatives, informed by these aforementioned models, as well as the five groupings of evaluation indicators gathered from the studies in our review (ie, impacts, participants’ reactions, perspective and behaviour changes, implementation, and continuous improvement). Because of our intent for the proposed framework to be applicable across all stages in the design, implementation and evaluation of physician wellness initiatives, we took inspiration from two common types of evaluation found in the literature: process (implementation activities identifying successes and areas of improvement) and outcomes (short-term and long-term impacts determining whether a programme has achieved its desired goals).133 141 By incorporating both in tandem, the proposed framework can more comprehensively assess the success of a programme before, during and after it has been put into action. Our framework was organised according to these evaluation types through the domains of ‘implementation’ for process evaluation and ‘impacts’ for outcome evaluation (see figure 2). Within each domain, we included subcategories that were synthesised based on the earlier evaluation indicator groupings in addition to the components of the five evaluation models. Overall, the framework’s dual domains and various subcategories highlight multiple pathways through which wellness initiatives can be evaluated. Graphical representation of a framework for evaluating physician wellness initiatives. Since a primary purpose for evaluation is to improve an initiative, cyclical arrows are embedded within the framework to illustrate that evaluation should be continuous and iterative. This feature aligns with the evaluation gap identified in our review, wherein authors highlighted the need for evaluation to span across short-term and long-term timeframes. Additionally, driving the framework are the concepts of fidelity (the extent to which the initiative is delivered as intended), flexibility (the ability to adapt an initiative to local or changing situations) and sustainability (the continued success of the initiative over time). As this framework was derived from the literature on physician wellness, the framework is intended to assess physician-specific initiatives; however, it may be adaptable to other healthcare professions given the universal nature of evaluation. By guiding users on what should be evaluated and facilitating revisions to the initiative based on identified gaps, the framework can help organisations integrate evaluation into the early stages of physician wellness programme planning, while also assessing the longitudinal impacts of these initiatives on physicians and the broader healthcare system. In doing so, the framework also aims to promote greater standardisation in the reporting of evaluations for wellness initiatives, which can further dialogue between institutions by allowing them to assess and compare strategies for improving physician well-being. Additionally, although aligning interventions with key drivers of physician burnout is essential to addressing the root causes of this issue, this process typically occurs earlier in programme development (eg, during needs assessments or intervention planning). For this reason, our framework is agnostic to aetiology not because these causes are unimportant, but because it is assumed that initiatives are already grounded in well-known drivers of burnout and have inherently addressed these factors at an upstream stage in the development process. Accordingly, our framework is intended to be applied after an initiative has been appropriately designed to address causes of physician burnout and is therefore situated in the third stage of the following sequence: (i) identified factor, (ii) targeted initiative and (iii) systematic evaluation. In the future, we intend to develop an implementation guide and pilot the framework by applying to a wellness initiative at an organisation to pinpoint gaps and make revisions with consultation from participants and stakeholders. The process will help to detail the elements expected under each domain and subcategories, to ensure the framework is practical and accessible for users. Implementation domain The implementation domain focuses on the usage of resources in implementing and delivering the initiative’s activities, as well as early results. Housed within this domain is a category called environmental variables, which include the elements of the organisation, system and broader environment that facilitate or act as a barrier for the initiative. For example, the organisation’s culture towards physician well-being may affect the extent of funding allocated to initiatives. Likewise, government policies such as mandates for working hours could spur adoption of initiatives. Implementation metrics refer to the direct, quantitative outputs from the initiative’s activities to inform ongoing development and delivery. A common example of this is programme costs. Alternatively, as a more specific example, an initiative to improve physician experience with EHR systems could be assessed by measuring the amount of time spent on the EHR system after hours. Experiences refer to participants’ holistic reaction to the initiative in terms of acceptability and relevancy, as well as broader impacts to participants’ lives. This may include physicians’ perceived usefulness of an initiative in addressing their wellness needs or impacts on a physician’s work-life balance. Impacts domain The impacts domain focuses on the results of the initiative and whether they meet stakeholders’ expectations post-implementation. Although the framework is targeted at physician wellness initiatives, impacts do not necessarily need to be limited to this population, particularly as it is noted that physician burnout has implications across the healthcare system. For example, an initiative to improve physician wellness may be interested in how quality of patient care is affected. Within this domain, individual-level outcomes are the changes in individuals’ knowledge, attitudes, skills and behaviour as a result of the initiative. An example of this could be whether an individual gains stress management skills from a workshop, and also their ability to apply those skills in practice. Organisation-level and system-level outcomes encompass the overarching results that the programme intends to achieve, such as improved physician retention and efficiency of practice. These outcomes also include unintended or negative outcomes, such as stigmatisation of participants. Limitations Over 85% of the studies reviewed originated from four countries (the USA, Australia, Canada and the UK), potentially limiting the generalisability of findings to other regions. Additionally, only five evaluation frameworks were identified, as the focus was on physician wellness, but broadening the scope to include frameworks from other fields (eg, nursing or education) may have yielded richer perspectives and dimensions to evaluation. A further limitation relates to the distinction between initiatives implemented within healthcare organisations and those delivered through educational or training contexts, such as residency programmes. Although residency training occurs within hospital settings, many wellness initiatives targeting residents are implemented through educational structures (eg, residency curricula or programme-level activities) rather than through healthcare organisations themselves. As this rapid review focused on initiatives implemented at the healthcare organisation level, some residency-based interventions may not have met our inclusion criteria. Conclusion While we identified evaluations of initiatives focusing on physician wellness that were diverse in scope, there was no standardised method for evaluation across the studies reviewed. Data collection for evaluations varied in their approaches and timeframes. A plethora of evaluation indicators was used, measuring deliverable components of the initiative, how participants experienced the initiative and the different outcomes of interest. We consolidated these insights to develop an evaluation framework specifically targeted towards wellness initiatives for physicians, given the population’s unique relationship with their institutions and the important burnout-related outcomes for the healthcare system. This framework can help evaluators identify areas for improvement by assessing the effectiveness of wellness initiatives through relevant indicators and outcomes. Given the range in how studies conducted their evaluation, a well-rounded evaluation should balance comprehensiveness and brevity; in other words, it considers the essentials of evaluation with what can be practically achieved in demanding healthcare settings. Supplementary Material Reviewer comments Author's manuscript Footnotes Contributors: IK and SK contributed equally to this paper and are joint first authors. TW and SS jointly supervised the work and share co-senior authorship. IK and SK contributed to data acquisition, data extraction and analysis and writing (original draft). BL and TW contributed to study conception and design, data analysis, and writing (review and editing). TT and SS contributed to study conception and design, and writing (review and editing). TR contributed to data acquisition. All authors approved of the final draft. IK is the guarantor. Funding: This work was supported by the Ontario Medical Foundation and Dr Bill (#0876037). Competing interests: None declared. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Provenance and peer review: Not commissioned; externally peer reviewed. Supplemental material: This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise. Data availability statement Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information. Data availability statement Data extraction of reviewed studies is available from the corresponding author (treena.wilkie@camh.ca) upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information. Ethics statements Patient consent for publication Not applicable. Ethics approval Not applicable. References Associated Data Supplementary Materials bmjopen-16-4-s001.pdf bmjopen-16-4-s002.pdf bmjopen-16-4-s003.pdf Reviewer comments Author's manuscript Data Availability Statement Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information. Data extraction of reviewed studies is available from the corresponding author (treena.wilkie@camh.ca) upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information.