Educational roles as a continuum of mentoring’s role in medicine – a systematic review and thematic analysis of educational studies from 2000 to 2018

Background Recent studies have gone to great lengths to differentiate mentoring from teaching, tutoring, role modelling, coaching and supervision in efforts to better understand mentoring processes. This review seeks to evaluate the notion that teaching, tutoring, role modelling, coaching and supervision may in fact all be part of the mentoring process. To evaluate this theory, this review scrutinizes current literature on teaching, tutoring, role modelling, coaching and supervision to evaluate their commonalities with prevailing concepts of novice mentoring. Methods A three staged approach is adopted to evaluate this premise. Stage one involves four systematic reviews on one-to-one learning interactions in teaching, tutoring, role modelling, coaching and supervision within Internal Medicine, published between 1st January 2000 and 31st December 2018. Braun and Clarke’s (2006) approach to thematic analysis was used to identify key elements within these approaches and facilitate comparisons between them. Stage two provides an updated view of one-to-one mentoring between a senior physician and a medical student or junior doctor to contextualise the discussion. Stage three infuses mentoring into the findings delineated in stage one. Results Seventeen thousand four hundred ninety-nine citations were reviewed, 235 full-text articles were reviewed, and 104 articles were thematically analysed. Four themes were identified – characteristics, processes, nature of relationship, and problems faced in each of the four educational roles. Conclusions Role modelling, teaching and tutoring, coaching and supervision lie within a mentoring spectrum of increasingly structured interactions, assisted by assessments, feedback and personalised support that culminate with a mentoring approach. Still requiring validation, these findings necessitate a reconceptualization of mentoring and changes to mentor training programs and how mentoring is assessed and supported.


Methods
To evaluate the notion that teaching, tutoring, role modelling, coaching and supervision may be a part of an overarching concept of mentoring, this study was made up of three stages. Stage 1 consists of systematic reviews of teaching, tutoring, role modelling, coaching and supervision carried out to provide better understanding of these processes. In acknowledgement of mentoring's, coaching's and supervision's context-dependent, approach-specific nature, studies were confined to educational accounts that involve one-to-one interactions between tutor and learner. The term tutor was used to encapsulate mentor, supervisor, teacher, role model and coach.
Stage 3 sought to determine similarities between teaching, tutoring, role modelling, coaching and supervision and mentoring.
To carry out the systematic reviews, Stage 1 adopted Braun and Clarke's (2006) approach to thematic analysis to identify key themes within teaching, tutoring, role modelling, coaching and supervision in medical education [17].
Stage 3 sought comparisons being made between novice mentoring and teaching, tutoring, role modelling, coaching and supervision to determine the overlap between each of these approaches.
Stage 1: thematic analysis of teaching, tutoring, role modelling, coaching and supervision Methodology A systematic review was proposed to explore the size and scope of available literature on assessing the impact of medical ethics education in published peer-reviewed literature [18][19][20][21][22]. This allowed for systematic extraction and synthesis of actionable and applicable information [23] whilst summarizing available literature [24,25] across a wide-range of pedagogies, assessment contents and practice settings [26][27][28][29][30]. Levac et al. (2010) [31]'s and Arksey and O'Malley (2005) [18]'s framework for systematic review was used to map "the key concepts underpinning a research area and the main sources and types of evidence available" [21] and "produce a profile of the existing literature in a topic area, creating a rich database of literature that can serve as a foundation" to inform practice and guide further research [19,32,33].
Stage a: identifying the research question Guided by two librarians from the medical libraries at Yong Loo Lin School of Medicine at National University Singapore and the National Cancer Centre Singapore and educational experts and clinicians from the Singapore General Hospital, the Division of Cancer Education at the National Cancer Centre Singapore and the Marie Curie Palliative Care Institute at the University of Liverpool (henceforth the advisory team), the 14-person research team (YR, JY, AH, KT, KP, NQ, RP, BT, AC, YP, RK, DT, SM, and LK) discussed prevailing concepts and practice surrounding issues, and practices surrounding teaching, tutoring, role modelling, coaching and supervision and identified the primary research question to be: 'what is known of teaching, tutoring, role modelling, coaching and supervision in Internal Medicine?'. The secondary questions were "what are the key elements of teaching, tutoring, role modelling, coaching and supervision in Internal Medicine?'. These questions were designed on the population, concept and context elements of the inclusion and exclusion criteria [34,35], using a PICOS format (Table 1).
Guided by the advisory team, the 14-person research team worked in teams of threes under the supervision of the senior researchers (LK, SM, DT, and RK) and supported by near peer mentors (YP and KT) to carry out independent searches of accounts of role modelling, teaching, tutoring, coaching and supervision published in the PubMed, Scopus, ERIC and Cochrane Database of Systematic Reviews. The searches were carried out between the 12th September 2017 and 18th October 2017. The respective search strategies are found in the PRISMA in Fig. 1 With all searches reviewed by the senior reviewers, the review process was extended, and additional searches were performed between 12th May 2019 and 24th April 2019 to review newly published literature from 1st January 2016 to 31st December 2018 for each of the learning approaches.
Focus was on accounts of role modelling, tutoring, teaching, coaching and supervision that clearly described one-on-one interactions between a clinician and a learner in Internal Medicine. Accounts of teaching, tutoring and role modelling that did not clearly state one-on-one interactions were excluded as it did not facilitate comparisons with mentoring, supervision and coaching. Accounts of teaching, tutoring, role modelling, coaching and supervision in clinical specialities not traditionally associated with Internal Medicine as defined by the World Health Organization's classification of healthcare workers, were also excluded to further focus this review [36]. Braun and Clarke's (2006) approach to thematic analysis [17] was used to circumnavigate the wide-range of research methodologies that made statistical pooling and analysis difficult [17] in the papers reviewed. The narrative produced was guided by the Best Evidence Medical Education (BEME) Collaboration guide [37] and the STORIES (Structured approach to the Reporting In healthcare education of Evidence Synthesis) statement [38].
Stage B: identifying relevant studies Guided by the advisory team, the research team developed individual search strategies for teaching, tutoring, role modelling, coaching and supervision and selected PubMed, Embase, PsycINFO, and ERIC databases for review. In keeping with Pham et al. (2014) [33]'s approach of ensuring a viable and sustainable research process, the research team confined the searches to articles published between 1 January 2000 and 31 December 2018 to account for prevailing manpower and time constraints faced by the team.
Stage C: selecting studies to be included in the review After the independent searches of the databases were combined employing the 'negotiated consensual validation' approach and a final list of article to be reviewed was determined, the 7-members of the research team (YR, JY, AH, KP, NQ, RP, BT) guided by the senior reviewers (SM, RK, DT and LK) and near peer mentors (KT and YP) independently screened the title and abstracts.
A consensus based approach employing the 'negotiated consensual validation' approach was reached on the final list of papers to be included for thematic analysis [39].
The PRISMA charts are attached below (Fig. 1).
Stage D: data characterization and analysis In the absence of a priori framework and a clear definition of role modelling, teaching and tutoring, coaching and supervision, Braun and Clarke's (2006) [17] approach to thematic analysis was adopted to identify consistencies across these approaches [2,14,15,17,[40][41][42][43][44]. Braun and Clarke's (2006) approach was used to create codes from the 'surface' meaning of the data. Semantic themes were identified from 'detail rich' codes focused upon the various aspects of the role modelling, teaching, tutoring, coaching and supervisory process [17]. Each of the 10 coded scripts from role modelling, teaching and tutoring, coaching and supervision were reviewed by the senior reviewers. The research team discussed and agreed upon a common coding framework and codebook using Sambunjak et al. (2010)'s "negotiated consensual validation" approach [45]. Working in teams of three, overseen by the senior reviewers (SM, RK, DT, and LK) and peer mentors (KT and YP), the reviewers carried out independent thematic analyses of all articles in each of the four topics using the codebook, with new codes discussed online and at faceto-face at reviewers' meetings [17,[46][47][48][49].
Stage E: collating, summarizing, and reporting the results From the 18,938 articles reviewed, 34 articles on role modelling, 9 articles on teaching and tutoring, 43 articles on coaching and 18 articles on supervision were identified. The four themes identified include characteristics, processes, nature of relationship, and problems of the four educational roles.
The narrative produced was guided by the Best Evidence Medical Education (BEME) Collaboration guide [37] and the STORIES (Structured approach to the Reporting In healthcare education of Evidence Synthesis) statement [50].

Characteristics of each of the four educational roles
Thematic analysis of the prevailing descriptions of role modelling, teaching, tutoring, coaching and supervision were carried out. Their characteristics and descriptions are highlighted in the table below (Table 2).
The distinctive aspects of each learning processes are highlighted in Fig. 2.

Nature of relationship Role Modelling
Role modelling is unpredictable and involves varying levels of interaction and communication [100]. Some interactions are purposive; built through sharing of professional and personal experiences and socializing [60,63,101,102] whilst others are entirely opportunistic.

Coaching
The relationship between the coach and trainee is focused upon learning a specific skill [72]. The complexity of the skill determines the duration of the relationship [72].
Coaching begins with the demonstration of the skills in a planned role modelling process, which tapers over time as goals were achieved and as trainees develop their ability to self-monitor and sustain their training [72].
It is debatable as to whether coaching provides psychological and emotional support [74,80,84]. Some commentators suggest that coaching relationships are transactional and focused upon professional improvement whilst others suggest the presence of evaluations within coaching interactions necessitate a trusting [74,80,84] and safe environment [73,90,91].
A comparison of the nature of relationships across various educational roles are found in Appendix 1 in Table 4.
Role modelling faces limited time for teaching [55,100,[107][108][109] and bedside tutorials [110] whilst coaching faces inadequate financial, administrative and assessment support that are not conducive of nurturing organizational culture to ensure protected time and recognition for coaches [80,111].
Supervision faces organizational issues that include a lack of consistent level of support and training [93], resource limitation and competing tensions between service and education demands [95,96]. A detailed account of these challenges is found in Appendix 2 in Table 5.

Drawing the findings together
Based on the data from the four systematic reviews, it is possible to proffer a clearer understanding of each of the approaches.

Role Modelling
Positive role modelling can be defined as "a process where a trainer consciously or unconsciously demonstrates positive or negative behaviours, actions or attitudes. The learner observes, weighs up and reflects upon these characteristics, skills and or behaviours upon their own practice/attitude/behaviour and emulates, experiments, and assimilates it into his/her own personal/professional identity. Positive role modelling is more impactful when it occurs in a trusting, professional relationship."

Coaching
Coaching can be defined as a "longitudinal professional relationship between an expert coach and a trainee focused upon mastery of a clearly defined, measurable and achievable skill that is that the trainee or training organization feels the trainee can improve upon. The relationship is built upon professional trust in a 'safe environment' that facilitates practice of the skill. The coach evaluates the performance, needs and abilities of the trainee, role models skills, encourages learning, provides specific individualized feedback and devises a plan to achieve the goals. The trainee is accountable for their training and responsible for self-monitoring." [73][74][75][76][77][78][79][80][81][82][83][84][85][86][88][89][90][91] Supervision Supervision is an "individualized, focused, goal-specific, time-limited and context-sensitive clinical training process by a senior clinician aimed at assessing and improving particular gaps and weaknesses in the clinical care and patient safety by trainees by providing them with oversight, guidance and feedback and holding trainees up and accountable to established clinical standards and codes of practice. This process will utilize coaching and role modelling to meet its goals [92][93][94][95][96][97][98][99].
Comparing the findings of the four systematic reviews, there are a number of key insights and similarities that may be discerned. These commonalities lay the foundation for a collective perspective of the four educational approaches. These features are shown in Table 3 and the characteristics of each approach is shown in Fig. 3.
The data would suggest that the more relevant aspects of role modelling appears to be contained within teaching and tutoring, which in turn appears to be subsumed by coaching. Supervision appears to contain features of role modelling, teaching and tutoring, and coaching.
Viewed figuratively as concentric rings, role modelling would be at the centre of the rings, enclosed by teaching and tutoring, then coaching and finally supervision as the outer most ring.

Stage 2 mapping mentoring practice
To determine mentoring's relationship with the concentric rings featured in Fig. 3, Stage 2 will provide a summary of prevailing concepts of mentoring drawn from two recent systematic scoping reviews of novice mentoring and a recent study of mentoring experiences within a novice mentoring program. Sng   Evidencing these findings and forwarding new insights of novice mentoring, Krishna et al. (2019) [3]'s study of mentoring experiences in a novice mentoring program also unearthed new aspects to mentoring. These include 1. mentoring's competency-based stages of development that requires mentees to achieve basic competencies at each stage of the mentoring process before progressing to the next stage. 2. progress through the various stages of mentoring requires effective communication, timely and appropriate assessments and support appropriate balancing between consistency and structure.
3. oversight and support of the mentoring process depends upon the host organization and well-trained and supported mentors.

Stage 3 mentoring Spectrum
Mentoring's use of personalized holistic and longitudinal support throughout the mentoring process would require mentees to be taught, and provided with guidance as they apply their knowledge and skills, be assessed and provided with feedback and then re-evaluated before progressing to the next stage of the mentoring process. At each stage of the mentoring process which Krishna et al. (2019) [3] describe as "'circumscribed sequential projects' with 'specific goals and competency requirements'" , it is likely that mentors will employ role modelling, teaching and tutoring, coaching and supervision to support the mentee and the evolving mentoring relationship. This would see mentoring encompassing supervision's role and occupying the outer most ring in Fig. 4. This suggests that role modelling, teaching, tutoring, coaching and supervision lie within a mentoring spectrum Fig. 3 The relationship between the four approaches (Fig. 4). The mentoring spectrum describes a range of educational practices contained under the aegis of mentoring beginning with role modelling on the left side of the spectrum and mentoring on the right. Beginning with role modelling, there is progressively complex interactions culminating with personalized attention in role modelling behaviours, attitudes and practices, teaching new skills and knowledge, coaching individual learner's on different aspects of the skills they need, appraising their progress and providing feedback as they are supervised to complete their immediate goals within the project. Separating supervision from the more complex relationships seen in mentoring is the provision of personalized, timely, holistic and longitudinal support and the adaptation of the mentoring approach to accommodate the mentee's needs, goals, circumstance and abilities.
Other features that evidence the notion of a mentoring spectrum include

Motivation of learners
All educational approaches are reliant upon the learner's ability to observe, discern gaps in their ability, learn, reflect, weigh up considerations, be open to feedback and be accountable for their own learning. However, the approaches rely on increasing learner motivation moving from left to right of the mentoring spectrum.

Learning Relationship
Moving from left to right along the mentoring spectrum also highlights increasing planning and structuring of the mentoring process. Improved structuring of educational interactions better supports learning relationships and nurtures more holistic and personalised educational relationships. Better learning relationships also facilitate better outcomes.
Learning relationships also become more interactive moving from left to right in the mentoring spectrum. In role modelling, learners may not have an educational relationship with the tutor whilst learning relationships in mentoring are dynamic and enduring [2,92,[116][117][118][119].

Nurturing learning environment
Building a learning relationship relies on the learning environment and moving from left to right of the mentoring spectrum sees learning environments becoming increasingly important to the quality and nature of the learning relationship. These learning environments also become more individualised and serve to nurture particular learning relationships within the larger educational environment. This is especially evident in supervision and mentoring [90,91].

Learning assessment
Assessments also play an increasing role moving from left to right of the mentoring spectrum. These assessments must be timely, appropriate and personalised and accompanied by open and frank discussions and personalized, appropriate, specific, timely, holistic, accessible and longitudinal feedback and support [2,[116][117][118][119]. The presence of regular appraisals also reiterate the importance of individualized and safe educational environment [90,91].
The impact of the mentoring spectrum An overarching mentoring spectrum combining role modelling, teaching and tutoring, coaching, supervision, and mentoring has wide ramification upon how these educational approaches are employed.

The implication upon mentor training is significant.
Acknowledging the roles to be played within the mentoring spectrum highlights the need for mentors to be trained in all these educational approaches. 2. The unplanned and unconscious nature of role modelling and the need for balance between personalising and consistency within the mentoring approach both highlight the need for clear standards of practice, codes of conduct and practice guidelines (henceforth Codes of Practice or CoP). There must also be opportunities for mentee and mentors to align expectations and accept their responsibilities and roles and for mentees and mentors to be briefed on the prevailing goals and timelines of their respective educational projects and processes. 3. The learners and tutors must also be appropriately matched to ensure that they have complementary working styles, learning approaches and personalities, goals and abilities [120,121]. This will help build better educational interactions.
The implications of the mentoring spectrum upon mentoring practice is vast and includes requiring i. mentors-in-training to be trained and skilled on all these educational approaches and be mentored when applying these skill sets and competencies for each of the educational roles. ii. mentors and mentees to be briefed on CoPs, expectations on roles, responsibilities and expectations and effective oversight, assessment, and support provided by the host organization. iii. robust, longitudinal and holistic assessment processes in light of the changing nature of the mentoring process and the mentor's roles and the presence of evolving mentoring relationships and different stages of the various aspects within the mentoring spectrum. iv. the host organization to take an active role in overseeing and providing personalized, appropriate, specific, timely, holistic, accessible and longitudinal financial and administrative support in running and overseeing the mentoring process given the diverse processes within the mentoring spectrum [2,[116][117][118][119]. v. that the mentoring process is sufficiently structured to accommodate for the inevitable changes in the mentoring process without breaching the CoP. vi. the need for a safe and nurturing working environment that will nurture trusting and enduring mentoring relationships that will not only enhance better role modelling when the mentee has established ties with the mentor but also facilitate discussions that extend beyond professional issues which will allow the provision of holistic support. vii. the need for an open and safe mentoring culture that allows open discussions, constructive feedback and frank discussions.

Limitation
This review posits that these practices are interrelated is based on a number of novel yet unproven assumptions. Selecting only four of the many educational roles also limits the scope of understanding of the entire spectrum of educational roles in mentoring. In addition, the practices described in this review focus specific education settings, and draw from a particular definition of role modelling, teaching and tutoring supervision and coaching that may not be applicable in other education settings. Within the context of role modelling for example, there is no consideration of negative role modelling which limits the validity of the conclusions reached. In addition, many of the papers contextualized within the European and American healthcare system and training programs, limit their applicability to other educational and healthcare systems.

Conclusions
The findings of this review not only suggest a new way of conceptualizing mentoring but also highlights the need for further study into the matching, pre-mentoring, mentoring relationship, mentoring evaluations, mentoring structure, the mentoring environment and mentoring culture. This theoretical concept though supported by data from novice mentoring processes will still need to be carefully studied and validated. One key area for further study must be the manner that mentees, mentors and the host organization interact (mentoring dynamics) given its influence upon all processes within the mentoring spectrum. Similarly, important is the design of effective assessment tools and policing of the mentoring process and the mentoring environment. However, we are confident that this new concept of mentoring will enhance the mentoring process and mentoring outcomes as medical education strives toward personalized medical education.  Acquisition of standardized knowledge and skills, training for clinical competency [65][66][67][68][69][70][71], guided by formal teaching structure [66,69] Ensure trainees attain a minimum standard for safe practice with a focus on patient safety [93,96,98,99,104,122,123], good patient care [92,93] and clinical conduct [93,95,99,124]: -Provision of effective training [92,93,96,99,104,123] and monitoring, -Personalized supervisor feedback [94,99] -Gradual independence of trainees [124] towards their professional growth and development [92,93,122] Maximize the trainee's potential in a highly specific skill [82,84], involving complex objectives such as: -Communication skills [75, 77- [52,58,63] 3. Poor self-awareness of trainees [59,64,100,107,110], prone to emulating negative behaviours without careful judgement and reflection [59,64,100,107,110] 1. Poor program structure with inconsistent teaching guidelines [65], teaching qualities [69] and a lack of formal structure of tutoring [106] 2. Individualized learner's needs are unmet due to inability to accommodate to variable student personality, knowledge and skills [65,66,68], and dissonance in teacher-student learning needs [70] 3. Inaccurate program evaluation Self-rated outcomes and teaching rates are not predictive of students' performance [67,68]. There is a paucity of objective measure of learning behavioural competencies [68,70].
1. Inaccurate program evaluation and poor program structure [93,122] due to diverse perceptions of supervision practices [92,98,99,104,123,130], inaccurate assessment of trainee's needs and skills [93,104,122], lack of consistent, validated and objective outcome measures [92,99,104] 2. Lack of supervisor training with difficulty relating to the learners, and meeting their specific needs [92,122]. 3. Supervisor burnout with lack of protected time, interest, and presence of competing commitments [122]. 4. Suboptimal learning environment [92,99,123,129,130] with lack of supervisory feedback [132], fear of supervisory judgement, or loss of autonomy over learning [95] 1. Unsupportive coaching environment with conflicting educational roles as a coach, teacher, guide, and evaluator, deterring trainees from being genuine with their concerns [90].
Ms Yaazhini Renganathan is a third year medical student from the Yong Loo Lin School of Medicine, at the National University of Singapore. Email: yaazhini0206@gmail.com Mr Tay Kuang Teck is a fifth year medical student from the Yong Loo Lin School of Medicine, at the National University of Singapore. Email: taykuangteck@u. nus.edu Dr Benjamin Tan Jia Xing, MBBS, is a medical officer at the Singapore General Hospital, Singapore. Email: benjamin.tanjx@mohh.com.sg Ms Chong Jia Yan is a third year medical student from the Yong Loo Lin School of Medicine, at the National University of Singapore. Email: c.jiayan8@gmail. com Ms Ching Ann Hui is a third year medical student from the Yong Loo Lin School of Medicine, at the National University of Singapore. Email: annhui. ching@gmail.com Mr Kishore Prakash is a fifth year medical student from the Yong Loo Lin School of Medicine, at the National University of Singapore. Email: kishore@u.nus.edu Mr Nicholas Quek Wei Sheng is a fifth year medical student from the Yong Loo Lin School of Medicine, at the National University of Singapore. Email: nicholasquekws@u.nus.edu Dr Rachel Peh Huidi, MBBS, is a medical officer at the National Cancer Centre, Singapore. Email: rachel.peh@mohh.com.sg Ms Annelissa Chin Mien Chew is a senior librarian at the Medical Library, National University of Singapore Libraries, National University of Singapore, Singapore. Email: annelissa_chin@nus.edu.sg Professor David Taylor