Leadership Training in Graduate Medical Education: A Systematic Review

There have been numerous calls to increase leadership development for physicians in training.^1–3  Observational data have suggested that patient outcomes are improved with physician leadership.⁴  Some evidence also suggested residents are not prepared for informal or formal leadership roles following graduation.^1,5  Without formal training, physician leadership development can occur through “accidental leadership.”^1,6,7 

The Accreditation Council for Graduate Medical Education (ACGME) identified practice-based learning and improvement, interpersonal and communication skills, and systems-based practice as core competencies, all of which contain elements of leadership competencies.⁸  The CanMEDS framework used by the Royal College of Physicians and Surgeons of Canada renamed the role of “manager” to “leader.”⁹  The Academy of Medical Royal Colleges developed a Medical Leadership Competency Framework (MLCF) composed of 5 categories.¹⁰  These examples illustrate the support of graduate medical education (GME) accrediting bodies for leadership training for residents.

There are numerous resources for leadership development,^11,12  but the literature specific to undergraduate medical education and GME is limited.^13,14  Frich et al¹⁴  identified 45 physician leadership programs, including 26 aimed at individuals in GME. They reported an impact on knowledge, but few studies explored behaviors or higher-level outcomes, and noted deficits in interprofessional and experiential learning methods. Our review updates and expands the work of Frich et al and aims to identify common elements, best practices, and current gaps in GME leadership curricula.

We conducted a systematic review using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) standards and Association for Medical Education in Europe (AMEE) Guide 94.^15,16  Three databases were searched by a professional medical librarian (S.C.) in August 2014 for relevant English-language studies: MEDLINE, ERIC, and EMBASE. MedEdPORTAL was searched for relevant curricula. Literature search strategies used a combination of subject headings and key words relating to leadership, training, and GME. An updated search was completed in October 2015. Reference lists of articles selected for full-text review were hand-searched for additional articles.

Eligible manuscripts had to be full-length, peer-reviewed publications or MedEdPORTAL resources outlining programs to teach leadership in GME. As multiple definitions of leadership exist, broad inclusion criteria were used to maximize examples of leadership curricula. Articles were excluded if they solely addressed professionalism or teamwork in a particular setting (eg, running codes or operating room teamwork). Due to our interest in generalizable leadership curricula, we excluded curricula unique to specific settings.

A total of 3413 abstracts were independently reviewed by 2 authors (B.S. and J.D.H.; see the figure). Discordant abstract decisions were reconciled by the reviewers. We retrieved 201 articles for full-text review, and we deemed 52 eligible for inclusion (κ = 0.921, 95% confidence interval 0.853–0.989).

Two reviewers (B.S. and J.D.H.) independently extracted data from the 52 articles^7,17–67  using a standard data collection form that included specialty, training setting, trainee level, curriculum length, methods of instruction, use of mentors or coaches, experiential projects, educational theory, and MLCF competencies (demonstrating personal qualities, working with others, managing services, improving services, and setting direction).¹⁰  The quality of each article was scored from 1 (no clear conclusions) to 5 (results are unequivocal) based on Best Evidence in Medical Education (BEME) Guide No. 13.⁶⁸  Modified Kirkpatrick outcome levels (table 1) were used to determine program effectiveness.^69,70  Disagreement was uncommon, and it was reconciled using joint discussion with a third party (A.B.).

Three authors (B.S., A.B., and J.D.H.) reviewed the data and determined frequencies for the specific curricular components. We independently identified common elements of leadership curricula and their evaluations, and we summarized them through a collaborative process, using consensus to arbitrate areas of disagreement.

Of the 52 publications between 1991 and 2015, 34 (65%) were published in the last 5 years.^7,17–67  Key findings are presented in table 1, with specific characteristics and overview findings presented in tables 2 and 3 (and online supplemental material), respectively. Only 11 studies (21%) were deemed to have a BEME quality score 4 or 5. A majority (58%, 30 of 52) included primary care programs (family medicine, internal medicine, pediatrics, psychiatry); surgical (35%, 18 of 52) and subspecialty (33%, 17 of 52) programs were represented less frequently. Only 7 of 52 programs (13%) included multiple disciplines,^{7,20,31,33,40,47,65}  and 1 study was interprofessional.³⁸  Eight curricula (15%) were designed solely for chief residents.^{21,28,29,47,49,50,59,65} 

The majority of curricula reported using classroom-based learning (83%, 43 of 52) and small groups (69%, 36 of 52) to deliver the educational content. Of the 52 programs, 44 (85%) identified faculty as teachers, and 29 (56%) used outside consultants. Program length varied from isolated experiences (23%, 12 of 52) to serial lessons over a defined period (37%, 19 of 52) to longitudinal programs (40%, 21 of 52). Three programs (6%) required an extension in training.^7,22,33  Teamwork and models of leadership were the most common content (table 1).

The learning theory to design the curricula was reported in 7 articles (13%),^{19,28,39,42,48,56,64}  and 51 (98%) identified the pedagogic approach. The majority of leadership development programs used 3 or more strategies. This included 24 programs (46%) that described a project-based component.

Multiple assessment methods were used (tables 1 and 3), with survey-based methods being the most common (postcurriculum survey in 20 [38%]; pre/post survey in 16 [31%]). Five studies (10%) used pretests and posttests to assess knowledge. Mixed and qualitative methods were used in 9 studies (17%) and 6 studies (12%), respectively. Only 1 study used a control group to assess effectiveness.

Kirkpatrick effectiveness scores ranged from 1 to 3A, with 15 of 52 (29%) reporting level 4. A total of 28 articles (54%) reported learner satisfaction and 26 (50%) reported a subjective increase in knowledge. Objective evidence of learner knowledge and behavior change was reported in 13% (7 of 52) and 15% (8 of 52) of articles, respectively. The majority of Kirkpatrick level 4 scores were related to project completion.

Interest in formalized GME leadership curricula appears to be expanding, as our study found that 65% of articles were published after 2010. The majority of publications were from primary care specialties, and a variety of teaching modalities (lectures, conferences, and projects) were used. Curriculum evaluation was limited for most studies. The overall low quality of the articles as indicated by the BEME scores makes the identification of program elements that should be used in future curricula challenging. Available program resources and contexts, rather than applied theoretical constructs, appear to dominate the design and content of leadership curricula in residency.^1,12  Based on our analysis, we provide insights for developing and evaluating future curricula (table 4).

Most GME leadership curricula were published by primary care specialties (table 2). Surgical specialties and subspecialties had fewer articles, despite calls from leaders in the surgical community for leadership training.⁷¹  Pettit et al⁵⁵  demonstrated improved neurological surgery resident knowledge of leadership and an appreciation that leadership training was beneficial to developing physician leaders. A recent article examining the impact of leadership on adverse event reporting further highlighted the need for surgeon leadership to improve patient safety.⁷² 

Few programs reported interdisciplinary or interprofessional training, which is consistent with the report by Frich et al.^{7,20,31,33,38,40,47,65}  Interprofessional education may improve collaboration and team-based care, but it has yet to make it into leadership curricula.⁷³  Frich et al¹⁴  suggested that programs avoid multidisciplinary curricula to “foster a nonthreatening participatory and exploratory environment.” GME leaders should consider how multiple professions could participate in leadership development to prevent redundant training and facilitate interprofessional relationships.

Small group teaching, mentoring, coaching, and project work appear to be the most effective approaches for leadership training, and small group teaching was the favored teaching method for internal medicine residents.⁷⁴  Multiple modalities were used in studies with the highest effectiveness scores, consistent with the finding from Frich et al¹⁴  that the majority of programs used 2 or more methods. Steinert et al¹²  recommended using a variety of teaching methods, including experiential learning, reflective practice, projects, mentoring, and coaching, for teaching faculty leadership, and they highlighted that many programs attempted to match methods with objectives. Ultimately, methods should be selected based on desired educational outcome, available resources, and learner preferences.

Steinert et al¹²  also emphasized the importance of the adult learning theory in developing leadership curricula, yet only a minority of studies reported the learning theory used in program design. The learning theory could mold the curriculum in a pedagogically rigorous manner, allowing learners to practice their learning in the workplace.¹² 

Utilizing a leadership framework, such as the MLCF, which starts with personal qualities and moves to leadership that affects the system,¹⁰  may assist with curriculum development. Learning about oneself is an important component of leadership development,^1,12  and it was present in 41 (79%) of the articles in this review.

Improving personal leadership traits can be accomplished through reflective writing, self-assessments, and 360-degree evaluations.^1,12  Despite the importance of these methods for leadership development, they were rarely used, and effective and standardized tools were lacking or expensive.

Online learning was reported in only 2 articles (4%) and may be an area of growing focus.^12,22,58  Advantages of online training include asynchronous use, ease of assessment, better standardization, and scalability, yet online learning needs to be balanced with group discussion and reflection.

Reported content of leadership curricula was consistent with the work of Frich et al¹⁴  (table 1). An article from emergency medicine reported on 59 leadership competencies,⁷⁵  and a challenge for programs may be distilling a manageable list of core topics that cover key leadership competencies. Several programs used leadership training to augment ongoing clinical activities, such as quality improvement, resident-as-teacher programs, or running a team.

While the optimum time to introduce leadership training was not addressed in the studies included, our review demonstrates that longitudinal or serial sessions are more effective than a single training event. Learners reported satisfaction with experiential learning opportunities that empowered them to apply leadership skills.^{24,29,45,47,65}  Examples include team management, leading peers, small group facilitation, and addressing topics such as conflict resolution, feedback, managerial skills, and leadership styles.^45,49  Synchronizing leadership education with these experiences is consistent with adult learning theory.^12,76 

Residency programs are well-suited for longitudinal curricula and afford an opportunity to practice skills. Training for interns may focus on personal and interpersonal skills, while programs for more senior trainees may be geared toward team building and system change.^7,45,46  Change management, negotiation, and creating a vision could be taught when residents are conducting quality improvement projects. Many longitudinal leadership programs combine didactic or in-person learning with longitudinal projects.¹² 

In our review, 46% of curricula used projects, comparable to the findings by Frich et al.¹⁴  Project learning was more prevalent in high-quality studies, allowing trainees to apply leadership skills, and it was noted to be a “powerful motivational tool that enhanced accountability.”¹²  Quality improvement projects for trainees can be augmented with didactics, reflective writing, and small group discussions, and can afford opportunities to work in multidisciplinary and interprofessional teams.^1,12,14 

Mentorship was more commonly found in higher-quality studies, and it capitalizes on adult learning theory, allowing for formative feedback to learners. Many studies paired mentoring with projects to help facilitate completion, along with explicitly teaching leadership principles.^{7,22,26,27,32,33,35,36,40,42,44,47,51–54,56,60,66}  Mentorship also provided opportunities for relationships to continue after the formal curriculum ended.¹  Despite the appeal of mentoring, the availability of adequately trained faculty may be a challenge for training programs.¹² 

Few studies have addressed the faculty in leadership programs. Our review showed that program directors, departmental leaders, and hospital leaders comprised the majority of faculty. Many sessions used small group discussion. Thus, in addition to content expertise, facilitation skills are a desired competency for faculty. External consultants were used in the majority of programs, and they may provide greater depth to a program. For universities that have leadership programs outside of medicine, there may be opportunities to expand these programs to teaching leadership to trainees.

The overall low quality of studies was due to 2 major factors: (1) insufficient details about curricular design, and (2) low-level outcomes assessments, with most outcomes limited to learner satisfaction and self-reported knowledge.

While surveys and knowledge assessments are easy to obtain and may be useful in program feedback, these approaches fail to capture the full effectiveness of leadership programs. More comprehensive data (eg, evaluations from supervisors and peers) could better define Kirkpatrick levels 3B, 4A, and 4B. When examining the effect on systems, it is not clear that simply completing a project should be classified as level 4 effectiveness. This has a short-term effect on the system, but it may overestimate the effectiveness of the leadership program.

The timing of the program evaluation may be important for assessing effectiveness, with programs moving beyond immediate evaluation and examining longer-term outcomes to assess whether training has a lasting impact, as leadership skills may not be fully utilized until after residency. The lack of follow-up on leadership projects limits the ability to assess program effectiveness.¹²  Assessments should include immediate, intermediate, and long-term data to better understand effectiveness.⁷⁷ 

Qualitative methods provide insights into leadership experiences through the capture of narratives,⁷⁷  yet they were used in program evaluation in only 12% of articles in this review. Qualitative approaches facilitate understanding the complexity and nuances of leadership development, and they describe the meaning and significance of leadership development from learners' perspectives.^12,77 

This review has limitations. We did not capture unpublished leadership curricula, and we did not search the ABI/INFORM collection, where additional studies may have been published.^13,14 

Gaps exist in understanding the best ways to teach leadership and the value of leadership training. The overall quality of reported leadership curricula is low. However, the available evidence suggests that small group teaching, project-based learning, mentoring, and coaching are valuable components of leadership curricula. Longitudinal leadership curricula are more likely to be successful. Enhanced reporting of curricula design and examining higher-level educational outcomes would allow for more rigorous assessment of the value of leadership programs.