The delivery of high-quality health care is a complex endeavor at all institutions. Major challenges to achieving this goal include the existing gaps in quality of care, the complexity of caring for aging patient populations with chronic diseases, the debate about the appropriate use of new technologies and drugs, and the rapidly rising costs of care in a constrained economic time. Resolving these issues requires skilled, knowledgeable, and trusted leaders in health care organizations.1
Faculty at academic medical centers (AMCs) not only confront these challenges in the delivery of care but also have the additional responsibility of training the physicians of the future. In addition, excellent leadership is essential to establish strategic directions and build collaborations across multiple stakeholders in an academic health science enterprise.2,3 In these challenging times, many AMCs have initiated education programs to support and train the faculty who will take on these leadership roles now and in the future.4,5 Institutions are investing significant resources, both in time and money, to provide faculty with this training. For example, Drexel University College of Medicine sponsors “Executive Leadership in Academic Medicine,” which is designed to train women leaders. National medical organizations also recognize the growing demand for skilled leaders and have started to provide training programs for them. The Association of Professors of Medicine sponsors a program called “New Chairs and Emerging Leaders,” which is designed to teach incoming chairs of departments of medicine how to succeed at their new position. A burgeoning literature describes these programs and shares participants’ stories of their experiences as leaders.6–9
Despite the growing interest and the implementation of such leadership training programs, little is known about the impact of these efforts. What types of programs are most effective? What outcomes are they achieving? We conducted a systematic review of the literature on academic medicine leadership training programs to answer these questions and learn from the experiences of participants of existing programs.
We used the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement to guide our reporting here.10
An experienced information specialist, who is a colleague at our institution, developed our search strategy using medical subject headings and text words. In April 2011, we searched MEDLINE (OVID interface, 1948 to April 3, 2011), EMBASE (OVID interface, 1980 to April 3, 2011), CINAHL (EBSCO interface, 1994 to April 20, 2011), and the Cochrane Central Register of Controlled Trials (Issue 2, April 2011) for potentially relevant studies. We performed these searches without date restrictions but limited our results to studies published in English. For our full search strategy for MEDLINE, see Supplemental Digital Appendix 1, http://links.lww.com/ACADMED/A131. We modified this search strategy for the other databases (available on request). We supplemented this electronic search by scanning the reference lists of the studies we identified for additional relevant resources and by contacting a sample of the authors of the studies we had already identified (we found no additional studies as a result of this stage of our search strategy).
We included studies that reported on the implementation and evaluation of programs for physicians to develop leadership skills. Such programs could include any training initiatives for staff physicians targeting any type of leadership position, including those at academic health science/medical centers, universities (including undergraduate, graduate, postgraduate, and continuing education), or research institutions. Outcomes measures could include measures of academic promotion or other physician-level outcomes, such as job satisfaction, retention, absenteeism, leadership positions attained, and self-efficacy. We did not limit inclusion by study design (we included both qualitative and quantitative studies) or publication status. We excluded studies if no data (either qualitative or quantitative) were available to extract or if the studies were not published in English.
Study selection process
Two of us independently screened the titles and abstracts that we compiled during our literature search and the full text of any potentially relevant articles using the standardized eligibility criteria we described earlier. We resolved any conflicts by discussion and calculated our level of agreement on inclusion of articles using the kappa statistic.11
We developed our data abstraction form and pilot-tested it by each individually applying the form to three articles and discussing the results. We then clarified the form as necessary. Two of us independently abstracted data from each article, including study characteristics (e.g., study design, inclusion criteria, duration of follow-up), physician characteristics (e.g., mean age, percent gender, type of program leaders, years in practice), and outcomes (e.g., any outcomes were eligible including job satisfaction, retention, promotion, and collaborations). We resolved conflicts by discussion and, again, calculated level of agreement on data abstraction using the kappa statistic.11 We assessed the methodological quality of the included studies with the Critical Appraisal Skills Program worksheet for qualitative articles12 and the Newcastle–Ottawa Scale (NOS)13 for controlled before-and-after studies and case series. We modified the NOS for case series, excluding the elements about a control group.
Synthesis of included studies
Because of the significant heterogeneity between the study designs and outcomes examined in our included studies, we did not conduct a meta-analysis. Instead, we synthesized the results narratively and identified and discussed common themes as a team.
Our literature search initially resulted in 2,310 citations. After we eliminated duplicate articles and screened the titles and abstracts and full texts, we were left with 91 potentially relevant articles (see Figure 1). We excluded articles at the full-text level of screening when no data were available on the implementation or evaluation of the leadership program, such as when the authors provided only narrative descriptions of the program. Of these 91 articles, 11 met our inclusion criteria4,5,14–22; however, 2 of these publications described the same study,18,19 so we considered them together. We were left then with 10 unique studies and 11 articles. We reached good agreement amongst the review team at screening (kappa 0.85) and data abstraction (kappa 0.74).
Study and population characteristics
Of the 10 unique studies we identified, four were before-and-after case series,4,17,18,20 three were cross-sectional surveys,5,21,22 and three were controlled before-and-after studies14–16; one of the cross-sectional surveys21 also included an assessment of a sample of participants’ curricula vitae (CVs) after their completion of the leadership program (see Appendix 1). Three studies included a qualitative component.4,19,20 One study was completed in Denmark,14 one in Sweden,15 and the remainder in Canada and the United States.
All 10 studies together included 636 participants. All participants were staff physicians, but only 5 studies included details about their career stage and path.4,13–16 In those 5 studies, participants were at various stages of their careers and were following different career paths including leadership positions in education, research, and hospital administration (see Appendix 1). However, details on participants’ ages or years in practice were not available.
The leadership programs varied in their components and duration. Components included small-group seminars and workshops, executive coaching, mentorship, and team projects (see Appendix 1). They ranged from a one-time, two-day workshop4 to biweekly meetings spanning two years.21 Topics covered included leadership skills, mentorship, and strategies for leading teams (see Appendix 1).
The quantitative studies looked at the impact of leadership programs on a variety of physician-level outcomes, including attitudes (e.g., satisfaction with their skills), knowledge (e.g., knowledge about available resources), skills development (e.g., leadership skills), behaviors (e.g., assumption of leadership positions, absenteeism), and academic or job outcomes (e.g., promotion, retention) (see Appendix 2).4,5,14–18,20–22 The qualitative studies explored participants’ satisfaction with the program and their attitudes toward leadership.4,19,20 None of the studies reported data on the cost of the programs.
None of the 10 studies included details on participants’ age range, years in practice, or training; thus, it was unclear whether the participants were representative of academic faculty in general (see Appendix 3). Two of the controlled before-and-after studies lacked details about how participants were selected for the program or control group.14,15 All of the studies used self-reported questionnaires for most outcomes, and none described researchers’ blinding of outcomes assessment. Only two studies reported the use of validated questionnaires to assess outcomes14,15; the remainder did not report whether they did so. Two other studies included the assessment of participants’ CVs by reviewers to evaluate participants’ productivity and promotion, but these assessments were not blinded, and agreement across reviewers was not reported.16,21 Eight studies reported the percentage of participants with whom the researchers followed up after the program4,5,14–17,20,21; however, significantly more participants in the control group than the program group were lost during follow-up in two studies.14,16 Follow-up times varied across the studies and ranged from immediately after4 to 11 years after the program.21 Overall, we found that the 10 studies were at a high risk of bias according to our quality assessment (see Appendix 3).
Of the three studies that included a qualitative component, none provided sufficient details to allow us to conduct a quality assessment,4,19,20 and only one described the researchers’ analysis method.19
Impact of leadership programs: Quantitative data
Impact on attitudes.
Five studies (see Appendix 2) reported the impact of a leadership program on participants’ attitudes.4,5,15,17,22 von Vultée and Arnetz15 reported no significant differences between program and control group participants’ reports of well-being, mental energy, self-esteem, influence, authority, or efficiency, although they provided no quantitative data to support their claims. In a before-and-after case series, McDade and colleagues17 found that participation in a leadership program led to a greater acceptance of the demands of leadership (increase of 0.97 on a 7-point Likert scale 18 months after the program, P < .001). Participants in another before-and-after case series reported that the leadership program was most helpful in improving their attitudes toward time management and their ability to determine goals and priorities and foster effective interprofessional relationships, but the researchers provided no quantitative data to support the participants’ claims.4 Korschun and colleagues5 found that 98% (54/55) of participants reported an increased commitment to the institutional vision, and all expressed a willingness to participate in research, education, or clinical practice after completing the program. McAlearney and colleagues22 found that program participants reported being more motivated to get involved with professional activities within and outside their organizations (mean score of 3.7 on a 5-point Likert scale). These participants also reported having greater confidence in the strategic direction of their hospital (mean score of 4.1).
Impact on knowledge.
Four studies (see Appendix 2) reported the impact of a leadership program on participants’ knowledge.16,17,20,22 In one study, participation in a leadership program was associated with a greater knowledge of leadership theory than participation in the control group (P = .005) and with a trend for lower scores in participants’ confidence in their financial management knowledge.16 In a before-and-after case series, knowledge of career advancement sophistication (increase of 1.52 in composite score on a 7-point Likert scale, P < .001), leadership and organizational theory, and career building increased after participation in a leadership program (increase of 1.79 in self-reported composite score on a 7-point Likert scale, P < .001).17 In another before-and-after case series, all participants’ ratings of self-assessed knowledge and understanding of leadership (using a 5-point Likert scale) increased six months after the leadership program.20 In a cross-sectional survey, participants felt that they were more aware of the hospital resources available to facilitate their leadership roles (mean score of 4.1 on a 5-point Likert scale) after completing the program.22
Impact on skills.
Seven studies (see Appendix 2) described the change in participants’ skills as a result of participating in a leadership program, but most were self-reported assessments.5,14–17,20,22 Malling and colleagues14 found no significant difference between participants in the program and control groups according to their multisource feedback score (mean score 5.8 [standard deviation (SD) 0.6] versus 5.5 [0.5], respectively), which was provided by a human resources consultant. This score included technical, human, and administrative skills as well as citizenship behavior. von Vultée and Arnetz15 observed no significant difference in skills development scores (assessed using the Quality, Work Competence tool) between participants in the program and control groups. Day and colleagues16 found that program group participants scored higher than those in a control group in seven of eight leadership competency scores. Between the program and control groups, three of these self-reported scores were significantly different, including tolerance of the demands of leadership (P = .001) and understanding of leadership positioning (P = .008).16 In their before-and-after case series, McDade and colleagues17 reported that participation in a leadership program led to increased self-reported skills in financial management (increase of 1.91 in composite score on a 7-point Likert scale, P < .001), environmental scanning (increase of 1.47 in composite score, P < .001), communication (increase of 0.84 in composite score, P < .001), networking and coalition building (increase of 1.26 in composite score, P < .001), general leadership skills (increase of 0.95 in composite score, P < .001), assessment of personal strengths and weaknesses (increase of 1.28 in composite score, P < .001), and conflict management (increase of 1.26 in composite score, P < .001). In a before-and-after case series, McCurdy and colleagues20 found that participation in the program increased self-reported skills in self-management, managing others (such as in a research or project team), and managing others within a system (such as managing a group of individuals within a clinical portfolio in a hospital) six months after the program (P < .05 for each, using a 5-point Likert scale). Korschun and colleagues5 found that all survey respondents agreed that team leadership skills improved to some extent 18 months after enrolling in the leadership program. McAlearney and colleagues22 found that participants reported being more effective working in teams (mean score of 4.0 on a 5-point Likert scale) and in leading a team (mean score of 4.3 on a 5-point Likert scale) after completing a leadership program.
Impact on participants’ behaviors.
Six studies (see Appendix 2) described the impact of a leadership program on participants’ behaviors.5,15,16,18,21,22 von Vultée and Arnetz15 found that participants in their leadership program had taken fewer sick days one year after the program (mean 1.3 days [SD 0.4] versus mean 12.0 days [SD 14.9], P = .01) than those in the control group. Day and colleagues16 identified that participants in both the program and control groups were productive during the follow-up period but that participants in the program group had more peer-reviewed publications (3.5 per year versus 2.1 per year, P < .001). Similarly, both cohorts showed an increase in the number of leadership roles on national committees held by participants, but this increase was greater for participants in the program group (22% [baseline] to 62% [follow-up] versus 11% [baseline] to 19% [follow-up], P = .05). The researchers collected these data on publications and participation on committees by reviewing participants’ CVs. Osborn and DeWitt18 reported that 60% (18/30) of participants in a leadership program subsequently assumed a leadership (local or national) position, 43% (13/30) conducted a workshop, and 27% (8/30) obtained a peer-reviewed grant over a two-year follow-up period. In their survey, McAlearney and colleagues22 found that participants reported new and expanded leadership roles after participation in a leadership program (self-report score, mean of 4.0 on a 5-point Likert scale). Wilkerson and colleagues21 assessed participants’ CVs after their participation in a leadership program and found that 61% (43 of 71 participants who were still at their institution at the time of the study) assumed a new leadership role at their university, were appointed to an educational committee, or developed new teaching materials; 67% (30/45) of participants completed an education program, and 29% (13/45) were appointed to lead an educational committee. Korschun and colleagues5 found that 76% (42/55) of participants assumed additional responsibilities after the program, whereas 31% (17/55) remained at the same level of administrative responsibilities one to three years after completing the leadership program, although no baseline rates were available for comparison.
Impact on participants’ outcomes.
Three studies (see Appendix 2) described the impact of a leadership program on participants’ outcomes.5,16,18 Day and colleagues16 found that the percentage of program participants who advanced in their academic (48% versus 21%, P = .005) and hospital rank (30% versus 9%, P = .008) was greater than the percentage of participants in the control group. Osborn and DeWitt18 reported that 7% (2/30) of program participants were promoted according to a two-year follow-up. Korschun and colleagues5 reported that 15% (8/55) of participants were promoted within three years of completing the program, and 96% (53/55) were more likely to stay at their institution. However, Osborn and DeWitt18 and Korschun and colleagues5 provided no data for comparison.
Impact of leadership programs: Qualitative data
According to McAlearney and colleagues,22 participants in leadership programs reported that they were able to apply concepts from the program to their job and to change their leadership behavior. Participants in the study by Steinert and colleagues4 reported that the most useful elements of their two-day leadership program were lessons in time management, determining goals, and preparing meeting agendas. One year after the program, however, participants stated that they were less successful than they would like to be at saying no, delegating, adopting leadership styles, and evaluating meetings, and they suggested a “booster” intervention. McCurdy and colleagues20 reported that the themes of personal learning projects for program participants included personal changes, organizational changes, management issues, and interprofessional skills.
Although the business world has supported leadership training for decades, AMCs more recently have recognized the potential benefits of education for future leaders. AMCs are developing such programs and investing significant resources, including faculty time and money, to implement them. Yet, we found a remarkable paucity of literature that presents the benefits of such leadership training.
The existing published reports that we did identify, however, have important deficiencies in design. Most used self-reported data from participants and lacked objective, blinded outcomes assessments; only two studies used validated questionnaires for outcomes assessment.14,15 In addition, most studies assessed the benefits at the end of the leadership program rather than at a later time when participants potentially could have applied their newly acquired skills and knowledge in the work setting. These limitations impair our ability to learn from these early and significant investments in leadership training.
Despite these shortcomings, we can learn something from the published literature. Day and colleagues’16 study of a leadership program for orthopedic surgeons, for instance, included several strong design elements—the use of a control group, a follow-up across an average of 4.1 years, and an outcomes assessment based on the review of participants’ CVs. They demonstrated a positive impact on academic advancement and hospital leadership roles, both of which are important outcomes for AMCs.16 Two studies also demonstrated that participation in a leadership program had an impact on participants’ self-reported knowledge and skills related to relevant topics, like networking, coalition building, and environmental scanning,17 and success in obtaining grants and publishing papers.16 In addition, Day and colleagues16 demonstrated a difference between program participants and control group participants in attitudes toward leadership in general and tolerance for the challenges of leadership positions. Finally, the study of novel outcomes like sick leave taken15 or multisource feedback scores14 offer opportunities for us to understand the potential broad impact of these programs. Taken together, the findings of these studies indicate that leadership programs can have modest effects on outcomes that are of importance to AMCs.
Many AMCs have lacked women in senior leadership roles. For example, in 2000, only three women served as chairs of departments of medicine at U.S. or Canadian medical schools.23 Some AMCs have seen this paucity of women in leadership positions as a deficit, and two studies did focus exclusively on training women leaders.15,17 These studies by von Vultée and Arnetz15 and McDade and colleagues17 describe efforts specifically to design leadership programs for women who were selected by their deans or an equivalent because of their potential to become leaders at their AMCs. Typically, these programs included an emphasis on mentoring so that women could receive support and encouragement from other women already in these roles. However, we could not determine from these studies whether participation in these leadership programs will help to address the paucity of women leaders or what components of leadership training are most effective in this regard.
Not only were we surprised by the paucity of studies on leadership programs in general but also by the lack of qualitative studies specifically. We found only three studies that included qualitative results.4,19,20 Because developing leadership attitudes and skills is a personal journey for individuals, qualitative studies may be particularly useful in understanding the evolution of physicians’ career aspirations and goals over the course of training. Qualitative studies also may illuminate our understanding of the aspects of leadership education that physicians perceive as the most important or influential on their career development. In other fields, like education, qualitative methods more often have been used to study leadership; for example, Parker Palmer24 has led a variety of efforts, including publications and workshops entitled “Courage to Teach,” to explore the development of teachers. The business world also has used qualitative analysis to explore successful leadership styles.25
Our systematic review has several shortcomings. First, we limited our search to articles in English because we were only interested in programs that were conducted in English. Second, we focused on leadership programs in academic settings, thus excluding any studies that focused exclusively on non-university-affiliated hospital leadership programs, which we considered outside the scope of our review. Third, as with any systematic review, our results are dependent on the quality of the published studies that we found, which had significant methodological weaknesses as we have reported. Furthermore, the studies were very heterogeneous, with a broad variety of outcomes ranging from traditional academic metrics of success, such as publications and grants, to more novel ones, such as lost work days. Because few studies used the same metrics for looking at attitudes, knowledge, or skills, we were unable to conduct a meta-analysis or to assess publication bias. Finally, there are many tools that we could have used for quality assessment, including the Medical Education Research Study Quality Instrument (MERSQI)26 but we selected the NOS,13 which can be used for observational studies. Moreover, it is less helpful to provide a summary quality score that does not indicate to readers what the sources of bias were in the individual studies. Instead, we have provided the details on the quality assessment so that readers can judge the studies for themselves.27 Also of note, we did not exclude any studies on the basis of reported quality, to ensure transparency in our review.
Based on our findings, there are opportunities for future studies to build on our knowledge of effective leadership education. First, study designs should ensure that target populations are well defined (including physicians in different career paths), that interventions are clearly described (including the “dose,” “formulation,” and duration), and that outcomes are measured in a blinded fashion with appropriate metrics. These design features are feasible in studying leadership training; for example, the blinded assessment of CVs to determine whether participants have attained leadership roles is a reasonable and easily implemented outcomes measure. In addition, multisource feedback is well suited to the assessment of leadership behaviors from a variety of different stakeholder perspectives. Finally, researchers who integrate qualitative methods into their studies may learn what aspects of leadership training are most effective and what processes are most useful to participants.
As demands on AMCs to manage complex care in a constrained economic environment increase, effective leaders are essential. According to our findings, AMCs should continue to develop and implement leadership programs despite the significant costs to the institutions and individuals participating. High-quality studies will allow the medical community to share best practices and understand which programs are most effective.
Acknowledgments: The authors thank Laure Perrier for her assistance in completing the literature search, Andrea Tricco and Scott Wright for commenting on a previous draft of this report, and Desiree Chanderbhan for obtaining the articles.
Funding/Support: SES is supported by a Tier 1 Canada Research Chair.
Other disclosures: None.
Ethical approval: Not applicable.