Souba, Wiley MD, ScD, MBA; Notestine, Mark PhD; Way, David MEd; Lucey, Catherine MD; Yu, Lianbo PhD; Sedmak, Daniel MD
The roles and responsibilities of the clinical department chair in academic health centers (AHCs) are both complex and evolving. As noted by David Korn,1 former dean of Stanford Medical School, “Perhaps nowhere in the evolving AHC will the impact of change be more stressful and consequential than for the traditional clinical academic chair.” Chairs are required to provide leadership to a broad range of constituents across the medical center while overseeing the center's tripartite mission of research, patient care, and education. Although interrelated, these missions often compete for institutional resources and attention.
Contributing to the convoluted complexity of the clinical department chair's role is the organizational structure of the modern AHC, which is becoming increasingly matrixed. In many AHCs, the clinical department chairs report to the hospital chief executive officer (CEO) and to the medical school dean. Mallon and Corrice2 reported that teaching hospital CEOs in 2009 were more engaged in the recruitment of clinical department chairs than they were in 2001. Many hospital CEOs view this recruitment as an opportunity to coordinate strategic planning of the teaching hospital with that of the medical school to improve performance and to advance the center's tripartite mission. Ideally, deans and CEOs are aligned on institutional mission, vision, and values, but this is not always the case.
Yet, now more than ever, the need for cooperation and synergy between the academic and clinical enterprises is vital. As noted by Sorensen,3 “Unless we are more creative and tenacious in transcending university and hospital organizational structures, we run the risk of jeopardizing our collective future.” The transformation of AHCs into complex business enterprises where “clinical revenue and academic performance support each other by being strategically and tactically aligned”4 is increasingly the model, and there is little disagreement on the part of most medical school and university leaders that research and clinical success are synergistic and interdependent.5 Given the vital role of clinical department chairs in both hospital and medical school governance,6 it is essential that deans and teaching hospital CEOs work collaboratively in hiring and evaluating clinical department leaders.
We carried out the study described below to compare and contrast the perceptions and perspectives of medical school deans and their “partner” teaching hospital CEOs regarding the responsibilities and priorities of clinical department chairs in today's AHCs. Given the essentiality of the alliance between the medical school and the clinical enterprise, we maintain that alignment and collaboration between medical school deans and teaching hospital CEOs are vital to the health of this partnership.
We developed a survey instrument designed to measure the perceptions and perspectives of medical schools and their partner teaching hospital CEOs; the instrument queried 34 items related to clinical chair performance. The items fit into six categories: mission prioritization, management responsibilities, values, skill sets, barriers to success, and competitive differentiators. The assessment tool was reviewed by survey experts within the Center for Education and Scholarship at the Ohio State University College of Medicine. Topics covered by the survey included managing the primary mission areas of academic medicine; key leadership responsibilities, leadership values, and attributes of a clinical chair; common barriers to the chair's success; and the competitive advantage effective chairs contribute to the business enterprise. Most survey questions asked the participant to rank a set of options in order of importance. There were six categories of questions (e.g., leadership responsibilities) with three, five, or seven questions within each category. Deans and CEOs were also asked about their own roles and responsibilities with regard to recruiting and evaluating clinical chairs. Additionally, survey participants were asked about their current roles and positions, and for information about their institution.
The population of interest for this study consisted of the deans of the 126 U.S. medical schools accredited by the Liaison Committee on Medical Education at the time of this study (2009) and their clinical enterprise counterparts, the CEOs of the primary teaching hospitals at the AHCs where the medical schools were located. For those medical schools that do not own their teaching hospitals, the CEOs (or senior executives) of the primary teaching or affiliate hospitals were surveyed. In most cases, the clinical enterprise leader was easily identified as the president and/or CEO of the health system or primary affiliate training hospital. Otherwise, the individual who had the title that best represented the leader of the patient care enterprise was selected. Although the senior leaders of some hospitals did not carry the title of CEO (often being referred to as executive director), they were included in the CEO cohort in data analysis.
We surveyed the deans and CEOs using both hard copy surveys distributed through the U.S. Postal Service and electronic surveys disseminated through the SurveyMonkey Electronic Survey Service. The deans received the survey directly from the investigators at Ohio State University College of Medicine. Surveys to the CEOs were administered by the University HealthSystem Consortium, an organization for leaders of AHCs. A cover letter explained the purpose of the study and enlisted the recipient's participation. The study protocol was reviewed and approved by Ohio State University behavioral science research review board.
We mailed the first wave of postal surveys in July 2009. Mailing was coordinated so that postal surveys arrived at about the same time as electronic surveys, giving the recipient the choice of method with which to respond. Follow-up reminders were sent three times between August and October 2009 through the SurveyMonkey service and three times through the mail. Surveys were coded and tracked as they were returned. Tracking provided the investigators with the ability to follow up with nonrespondents, match deans with their CEO counterparts, and assess how well survey participants represented the population at large.
We merged electronic and paper survey responses into one data set and analyzed with SPSS for Windows 17.0. Respondent and nonrespondent demographic data were analyzed for representativeness using chi-square tests of proportions. Institutional demographic characteristics tested for bias include region of the country (Northeastern, Central, Southern, and Western), type of institution (public or private), faculty size (small, <300; medium, 300–700; large, 700–1,100; extra large, >1,100), and whether the university owned its teaching hospital.
To simplify the task of completing the survey and to reduce the amount of time required for survey completion, we asked participants to rank items (because it is easier to rank a list of five items than it is to provide a rating for each of those items individually). The tradeoff for using ranked items is that the resulting data are more complex and difficult to analyze and contain an element of dependency (because each of the items that is assigned a specific ranking is completely dependent on that ranking for its evaluation). We also wanted to be able to use all of the data, and not just those in which we received matching responses from both the dean and CEO from the same institution.
To evaluate the consistency in the ranks of each item and within each set of items for both deans and CEOs, we collaborated with the Center for Biostatistics at Ohio State University to develop a measure of agreement based on the Friedman rank-sum statistic.7 The agreement measure (officially named the agreement index) is defined as the proportion of variation in the uniformly distributed ranks not explained by the variation of observed ranks and can be interpreted similarly to a correlation coefficient: Positive values closer to 1 indicate more agreement in rankings; values closer to 0 (and negative values, because there is no range restriction) indicate less agreement. The agreement index measures within groups of CEOs or deans and not agreement across the two groups.
To address the statistical complexities associated with having multiple items within sets of questions in each of the six categories, paired and unpaired data, and missing responses within set categories, the linear mixed-model analysis8,9 was used. This method models the correlated data with unrestricted covariance between items within a set of questions. Because the sample size in our study is relatively large, the use of the linear mixed model for these rank data is asymptotically valid. Redundancy was handled by dropping off the last item. Global, or omnibus, tests were implemented for the set of questions in each of the six categories (test of category by group interaction) to justify post hoc testing at the specific item level. When the global test of the set was significant, post hoc tests on the individual items within the set were tested with t tests. All linear mixed-model analyses were performed using SAS version 9.2.
The rates of return for the pairs of individuals from the 126 institutions surveyed were as follows: For deans, 84/126 (66.7%), for CEOs, 57/126 (45.2%), and for matched pairs of deans and CEOs from the same institution, 36/126 (28.6%). A total of 104/126 (82.6%) institutions were represented. There were 96 items missing from the 2,856 items submitted by deans, and 10 items were missing from the 1,938 items submitted by CEOs. Chi-square tests of proportion indicated that the survey respondents were proportionally representative of the four regions of the United States (deans: χ2 = 0.69, df = 3, P = .88; CEOs: χ2 = 2.49, df = 3, P = .48), institutions of varying faculty size (deans: χ2 = 0.85, df = 3, P = .84; CEOs: χ2 = 7.64, df = 3, P = .54), and public versus private institutions (deans: χ2 = 2.87, df = 1, P = .90; CEOs: χ2 = 1.46, df = 1, P = .23).
The responding deans and CEOs were biased with regard to hospital ownership (deans: χ2 = 16.35, df = 1, P <.001; CEOs: χ2 = 8.62, df = 1, P < .01). Deans from universities that own their hospitals were underrepresented, whereas CEOs from universities that own their hospitals were overrepresented.
Interpreting the linear mixed model and test of agreement results
Results of the linear mixed model and tests of agreement are displayed in Table 1 and Figure 1. For the set of questions in each of the six categories (e.g., mission prioritization), the global test indicates whether the CEOs, as a group, assigned their ranks to the items within the category in the same manner as did the deans as a group. A significant global test (P < .01) tells us that the assigned pattern of ranks across the two groups is different. A significant post hoc comparison tells us which of the items within a set contributes most to explaining the significant global test results (P < .05). These can be interpreted as a test of the difference in mean ranks given by the two groups.
The difference in mean ranks between the CEOs and deans is shown in Table 1. A negative difference indicates that the CEOs ranked the item as more important than the deans did, and a positive difference indicates that deans ranked the item as more important than the CEOs did. The standard error is the standard error associated with the post hoc tests of the mean rank difference. The lower this number is, the more confidence we can place that a significant test result is not attributable to sampling error.
The graphical representations in Figure 1 show plots of the mean ranks of items determined to be significantly different through post hoc tests and show separately their distributions for CEOs and deans. The agreement index value can also be found in this figure. The agreement index corresponds to the spread of the distribution of mean ranks. A high, positive agreement index value indicates high agreement among survey respondents. An agreement index value closer to zero indicates low agreement.
The CEOs and deans were asked to rank the relative importance of the chairs' responsibilities in terms of the three conventional mission areas: education, patient care, and research. By computing the mean rank for each item and for each group separately and sequencing those mean ranks from small to large, the order of importance across the two groups is identical, with patient care being considered the most important, followed by education and then research. However, the global test using the linear mixed model of ranks shows that the deans and CEOs are significantly different in their rankings of these three items within the mission prioritization category.
In looking at Table 1, we see that the largest mean rank difference is for the patient care item. Although both groups, on average, ranked this item first, the −0.51 mean rank difference tells us that the CEOs did so much more commonly than the deans. The post hoc test of the mean rank difference is significant, telling us that the group of CEOs are significantly closer to a rank of 1 than are the group of deans. Figure 1 shows that the CEOs as a group almost unanimously ranked patient care as the number 1 item, whereas the deans had a distribution of responses between rank 1 and 2. Finally, by looking at the agreement index, we see that the CEOs were in almost complete agreement in their ranking of patient care as the top mission area priority (agreement index = 0.88), whereas the deans' agreement in their ranking of patient care as number 1 was more variable (agreement index = 0.23). In other words, despite the surface similarity between deans and CEOs on their ranks of the mission priorities, the post hoc tests of these items demonstrate significant differences between the two groups on two of the three mission priorities. On average, a significantly greater number of CEOs ranked patient care first, and a significantly greater number of deans ranked research first. Both groups ranked education about the same.
Survey participants were asked to indicate the most important leadership responsibilities of a successful clinical chair by ranking five items in this category. As indicated by the significant global test, the deans and CEOs mostly disagreed as to the most important responsibilities. Although both groups' top three choices were the same, the sequence of their mean ranks was different. The deans' top three responsibilities were faculty recruitment and retention, developing a strategic vision, and ensuring quality, in that order. The CEOs' sequence was developing a strategic vision, closely followed by ensuring quality, and then faculty recruitment and retention. The two groups were nearly aligned with the sequence of their mean ranks on the remaining responsibilities in this set, financial and operations management and leading cultural change, which they ranked 4 and 5, respectively.
Post hoc tests reveal that the primary contributor to the significant global test of the responsibilities set was the significant difference in mean ranks for ensuring quality. A significantly greater number of CEOs ranked this item closer to the top (closer to 1) than did the deans (mean ranks of 2.18 versus 2.94). As was the case with the mission areas, the CEOs also had a higher level of agreement in their ranking of ensuring quality than did the deans (agreement index: CEOs = 0.45; deans = 0.28).
Survey participants were asked to rank the leadership values they believed to be the most important for a clinical chair's success. The global test showed a significant difference between the rankings of the two groups. Deans' and CEOs' sequences of mean ranks were similar for trust and accountability. Both groups ranked these items identically as numbers 1 and 2. The CEOs prioritized the remaining items in the following order: teamwork, results-orientation, respect, customer service, and mentoring. The deans prioritized the items slightly differently: respect, results-orientation, teamwork, mentoring, and customer service.
Two key differences in the mean ranks between the two groups were the two groups' rankings of mentoring and results-orientation. Deans ranked mentoring significantly closer to the top of the importance scale than did the CEOs (mean ranks of 4.81 versus 6.00) but ranked results-orientation significantly closer to the bottom (mean ranks of 4.21 versus 3.64). Levels of agreement for mentoring were almost identical to the levels of agreement for ensuring quality from the leadership responsibilities set of questions (agreement index: CEOs = 0.45; deans = 0.27). Levels of agreement were low for both groups' rankings of results-orientation (agreement index: CEOs = −0.02; deans = 0.05).
The deans and CEOs were generally in agreement regarding the attributes and skills required of a successful chair, as indicated by nonsignificant differences between the global tests. The two groups agreed on the top two of the attributes/skills they look for in recruiting clinical chairs: past record of achieving goals and effective communication. They also agreed on the bottom two: ability to work in a matrix organization and formal business training. The groups disagreed on their rankings of demonstrated problem-solving capabilities and prior leadership experience, but the differences were not significant. The deans ranked the former close to the top and the latter as fifth, whereas the CEOs ranked the former fifth and the latter third.
Barriers to success
The CEOs' and deans' mean ranks were identically sequenced for four of seven barriers to success, leading to global tests whose results were not significantly different. The two groups ranked inability to hold people accountable as the primary and most important barrier to success. The CEOs ranked inability to work with budgeted resources as the second-greatest barrier to success, whereas the deans listed ineffective communication. The only barrier that the CEOs and deans seemed to disagree on was the inability to balance competing departmental and organizational priorities. The CEOs' mean rank put this item at number 2, whereas the deans' fell out at number 5.
Finally, survey takers were asked to rank the areas of chairs' performance that contribute most to providing the organization with a competitive advantage. The global test indicates a significant difference in the manner in which the CEOs and deans distributed their ranks. Both groups ranked high-quality patient care as the number 1 item that contributes to an institution's competitive advantage, yet the mean ranks across the two groups are significantly different, with the CEOs ranking this item higher on average than did the deans. Levels of agreement within groups for the high-quality patient care item were higher for CEOs (agreement index: CEOs = 0.78; deans = 0.31).
The deans believed that other competitive advantages include clinical and translational research and medical student education. They ranked these items significantly higher than did the CEOs. The CEOs ranked cutting-edge technology and therapeutics significantly higher than did the deans. Other data gained from the survey suggest that deans are more involved in the recruitment of clinical chairs than are CEOs. Both CEOs and deans strongly agreed that the clinical chair should be involved in developing the strategic plan for the department's clinical program in the health system. The data also suggest that the alignment and relationship between the responding deans and CEOs are positive. Seventy-one percent of both deans and CEOs (54/76 and 40/56, respectively) rated the alignment between themselves and their counterpart in their AHC as either “excellent” or “good.” Sixty-four of 77 deans (83%) and 45/57 CEOs (79%) rated the relationship between themselves and their counterpart in their AHC as either “excellent” or “good.”
Conflicts between leaders of teaching hospitals and medical schools have existed for many years.10–13 This tension can be partly explained by competing priorities and differing accountabilities, which underscores the importance today of shared goals and collaboration. A key bridge between the dean and the top hospital executive is the clinical department chair, a physician whose responsibilities increasingly encompass managing the complex interconnectedness of the clinical, teaching, and research domains.14,15 More than ever, support for the recruitment of clinical department chairs—the so-called “package”—includes resources from the hospital (health system) as well as from the dean. Ideally, deans and CEOs collaborate effectively in the recruitment and appointment of new chairs of clinical departments.
We designed this study to compare the views of medical school deans and teaching hospital CEOs on the characteristics and skill sets they believed were most critical for the success of a clinical department chair. Although some key differences were apparent, we observed many more similarities, suggesting that deans and CEOs are, for the most part, aligned on how clinical department chairs should spend their time, what it takes to be successful, and what the most common shortcomings are that lead to failure. This is a significant change from earlier studies10–13 and demonstrates a healthy evolution in how deans and CEOs view their partnership and the relationship of their priorities.
We recognize the essentiality and “critical inseparability” of the tripartite mission at a successful AHC. Nonetheless, in evaluating the performance of a clinical chair, we asked the deans and CEOs to rank how important it was for a chair to oversee each mission. Both groups ordered the mission priorities of the clinical chair identically: patient care first followed by education and research. However, the CEOs were more likely to prioritize patient care over the other two missions, whereas the deans ranked the three missions more equally but prioritized research more highly than did CEOs. This difference reflects a more singular focus of hospital leaders on the clinical enterprise than deans, who must balance the competing priorities inherent to supporting all three missions.
A past record of successfully achieving goals was the most important attribute both groups looked for in recruiting a new chair, followed by evidence of effective communication skills and a fit with the institutional culture. Formal business training was not judged to be essential, which suggests that deans and hospital executives believe that fiscal and managerial skills can be learned “on the job” more readily than social and emotional intelligence, or that many administrative duties can be delegated to other members of the team with specific expertise. Although experience matters, it is increasingly clear that leadership is not an innate capability; rather, chairs and others can learn to lead and, indeed, must learn to lead.16–18 We also maintain that having a dean and CEO accessible to serve as coaches and role models is very helpful.
Similarly, there was general agreement between groups regarding the barriers that prevented a chair's success, so-called derailers or Achilles heels.19 Both deans and CEOs felt that the failure of chairs to hold their faculty accountable was the primary barrier to performance, followed by poor execution and ineffective communication. Given the importance of accountability as a characteristic of a high-performing culture,20 we found it curious that both groups ranked leading culture change last in this category, because it has been reported to be a key predictor of organizational transformation.21–25
Interestingly, both groups ranked ineffective mentoring skills as the least important obstacle to success. This observation could suggest that deans and CEOs believe that the clinical chairs' mentoring capabilities are quite adequate (we doubt it); alternatively, we wonder whether this result reflects an understandable but somewhat alarming emphasis on short-term, more pressing issues (e.g., budget shortfalls, retention issues, extramural funding challenges) over the long-term payoffs of faculty mentoring and advancement programs, which require considerable time and effort.
The deans and hospital leaders differed in their opinions regarding the competitive edge opportunities that a clinical department chair should emphasize. CEOs clearly emphasized high-quality patient care and cutting-edge technology and therapeutics as primary to competing. These are differentiators that hospitals and clinics use to gain market share locally and regionally. Deans, on the other hand, ranked distinction in clinical and translational research and educating future physicians as providing a unique competitive advantage. These are offerings that give an AHC distinction and reputation at a national and international level.
Both chairs and deans felt that the establishment of trust was the single most important leadership value of a successful clinical chair. As pointed out by O'Toole26:
In essence, the leadership challenge is to provide the glue to cohere independent units in a world characterized by forces of entropy and fragmentation. Only one element has been identified as powerful enough to overcome those centripetal forces, and that is trust. And recent experience shows that such trust emanates from leadership based on shared purpose, shared vision, and especially, shared values.
As mentioned, accountability was also a highly ranked value along with teamwork—both of these leadership values work best when grounded in trust. Interestingly, customer service was ranked last or near last by both cadres as a key leadership value, perhaps indicating that deans and CEOs see customer service as more of a “front lines” or “in the trenches” responsibility. Because quality was so highly ranked in contrast to customer service, it may be that deans and hospital leaders characterize customer service and quality patient care as distinct although not separate. Given the recent Hospital Consumer Assessment of Health Providers and Systems initiative, which now tracks and reports patient satisfaction as well as quality, the line between quality, outcomes, and customer service is likely to become more blurred.
That good leadership is correlated with superior organizational performance only makes sense. A recent study of AHCs demonstrated that tighter alignment between chairs and medical school deans on core values and on leadership climate perceptions correlated with higher medical school and department standing with the National Institutes of Health and higher U.S. News and World Report medical school and hospital ranking.27 Furthermore, major disparities between deans and chairs indicating a misalignment of perceptions and values were associated with poorer institutional effectiveness in the clinical and academic missions. The study did not establish a cause and effect, but it did demonstrate that agreement on leadership values and climate perceptions predicted superior performance. The findings we report here corroborate the importance of leadership alignment and concurrence.
In summary, the current study, despite the moderate percentages of responses, strongly suggests that deans and senior hospital executives (CEOs) at our nation's AHCs generally agree on the requisite skills, values, and responsibilities that make for a successful clinical department chair. Hospital leaders understandably place more emphasis on the clinical mission, but they recognize the importance of scholarship and teaching; our sense is that they are genuinely interested in learning more about clinical translational science awards, tissue banks, technology commercialization, and clinical skills labs. Simultaneously, deans' responsibilities include a primary emphasis on ensuring that the clinical chairs are active and accountable players in leading a high-performing clinical enterprise. This alignment is critical to both the business success of the AHC and to the integration of the tripartite mission. Not only is mission integration essential from an efficiency and effectiveness perspective, but it will become even more important as physician–hospital employment models mature, as the medical home develops, as health care becomes fully innervated by a powerful electronic backbone, and as the science of health care delivery evolves. Continued healthy collaboration between deans, CEOs, chairs, and other AHC leaders will ensure that the clinical, research, and education missions are optimally integrated and aligned, which will secure our ability to improve people's lives and simultaneously meet the needs of patients, insurers, donors, and legislators.
The authors wish to thank David Jarjoura, PhD, director, Ohio State University Center for Biostatistics, for his helpful comments and his assistance with the statistical methods used in this manuscript. The authors also wish to thank Barbara A. Anason, vice president and chief marketing officer, University Health Systems Consortium, and Peggy Fancler and Irene M. Thompson, of the University Healthsystem Consortium, for their help in distributing the survey to the CEOs.