Academic health centers (AHCs) attain economic, quality-of-care, and social benefits by improving physicians' and patients' satisfaction. The financial costs associated with physician turnover at AHCs are high. Recruiting a new primary care physician who is emerging from a training program costs approximately $236,000,1 and most of the expense occurs during the learning phase and adaptation of the physician to the new practice setting. Table 1 shows that the average physician-departure rate at the University of Missouri—Columbia School of Medicine (MUHC) from 1991 through 1998 was 7.69%. Another study reported turnover in internal medicine faculty to be as high as 38%.2 Using $236,000 as a reference, the cost of our average annual turnover was $3,823,750, with a $6,440,000 peak in 1997. This estimate may be too low; another calculated the cost of replacing a single primary care physician to be $900,000.3 AHCs often have many specialists who use expensive technology. Therefore, the expense of turnover among specialty physicians may be even higher, although data on this are limited.
The loss of clinical faculty, especially in small academic divisions, results in additional responsibilities for those faculty members who remain. These responsibilities, such as increased “oncall,” more administrative duties, and extra committee and teaching assignments, are likely to foster discontent for the remaining faculty members.
In contrast with academic institutions where the morale among physicians is low (which are often urban), rural physicians are satisfied with doctor—patient relationships, clinical autonomy, the care they provide for medically needy patients, and life in a small community. They tend to dislike limited access to cultural activities and amenities, inadequate time away from work, and bureaucratic interference. Physicians' satisfaction with their communities and their opportunities to achieve professional goals predict longer retention only after controlling for other areas of satisfaction.4 A study of physical medicine and rehabilitation faculty showed better career opportunities, personal and family matters, low job satisfaction with administration, and not enough financial rewards to be the most important reasons for their leaving their institutions.5 Respondents also identified what their old institutions could have changed to induce them to stay. Fewer administrative or political problems and more protected time were ranked highly by both those who remained in academics and those who left an academic institution. However, academic physicians also asked for effective role models, and others wanted higher salaries and more support for research.
Because of the cost and the disruptions that physician turnovers cause in clinical and research programs, the beliefs of faculty that affect their job satisfaction and retention bear investigating.
A multidisciplinary mid-career leadership program at MUHC, the Millennium Health Care Professional Development Program (funded by the Aetna Quality of Care Research Fund), annually challenges participants to address the key problems of our AHC. In 1999, one of the challenges was to study physician retention and incentives at a time of great stress in AHCs. The authors led the study of this issue.
Surveys were sent to faculty who had left MUHC within eight years of the start of the study (1991–1998). We used forwarding e-mail addresses, phone numbers, and Internet searches to locate former faculty members. Multiple attempts were needed occasionally. Surveys were also sent to physicians recruited into non-leadership positions at MUHC during the same time interval. Surveys were provided only to those who agreed to participate. Administrative staff from the medical school prepared the list of arriving and departing faculty to yield turnover rates. We used the university's physician-referral hand-book to determine the total number of faculty for each department during the study period.
We used a standard statistical software package to analyze the data. Non-parametric methods were used to determine statistical differences with the parameters associated with “satisfied” and “unsatisfied” respondents. We grouped the compiled data by department and remark type.
The sample population consisted of all non-management and non-emeritus physician—faculty members from any clinical department who were hired and/or who voluntarily departed from the MUHC's hospital and clinics staff from 1991 through 1998 (n = 223). The departments represented (number of surveys) were anesthesiology and perioperative medicine (10), child health (7), family and community medicine (6), internal medicine (20), obstetrics and gynecology (6), orthopedic surgery (6), physical medicine and rehabilitation (5), surgery (15), psychiatry and neurology (7), radiology (8), pathology and anatomical sciences (6), ophthalmology (6), and anonymous (3).
Survey Method and Instrument Design
Questionnaires either were completed during a semi-structured interview process (81%) or were delivered and returned by mail (19%). The first section of the questionnaire contained pairs of five-point Likert-scaled questions asking the respondents to (1) rate the importance of 14 specific institutional and leadership factors affecting their job satisfaction and (2) rate their perceptions of the availability of those factors. These factors were “written evaluation of your manager (e.g., section head),” “disclosure of department or organization budget data to faculty,” “clear track to advancement,” “faculty involvement in department/division strategic planning,” “protected time (for research or personal use),” “clinical quality review/assurance,” “local community amenities (e.g., nice place to live),” “mentor availability/research support,” “equitable distribution of salary/resources,” “stipends for extra responsibilities,” “trust—communication with chair/division head,” “positive collegial interaction,” “institutional reputation/image,” and “institutional support of clinical/research interest.”
The difference between the importance score and the availability score represents the degree of improvement needed in that area to meet the needs of the respondent (we called this the “lacking” score). We added the lacking score to the importance score to yield an overall priority ranking. For example, an importance rating of 5 assigned to a certain factor and an availability rating of 1 assigned to that same factor resulted in an overall priority score of 9 (5 − 1 = 4; 4 + 5 = 9). This was done not only to show what was important to physicians but also to show what needed to be improved to enhance recruitment and retention.
The second section of the questionnaire was a single question asking the respondents to rate their overall level of satisfaction with the AHC and four open-ended questions asking them to describe positive and negative aspects of their employment experiences at the university. Last, the survey documented whether participants had left the university or were on a tenure track.
Approximately five to ten minutes were required to complete the questionnaires. The questionnaires were coded and linked to respondents' names to prevent duplication of entries into the database.
A total of 105 individuals (46%) responded. Of the respondents, 44 had left the University system. The factor “trust—communication with chair/division head” ranked highest in importance to the respondents (see Figure 1). However, with lacking scores added, “protected time (for research or personal use)” had the highest priority for all respondents despite the overall satisfaction rating. “Local community amenities (e.g., nice place to live)” was the only factor to receive no lacking points. Other factors with high overall priority scores were “clear track to advancement,” “equitable distribution of salary/resources,” “trust—communication with chair/division head,” “stipends for extra responsibilities,” and “mentor availability/research support.”
In response to section two of the questionnaire, the mean overall satisfaction scores (scale = 1 to 10) ranged from 3.8 to 7.6 depending on the respondents' clinical departments. This score correlated inversely (p = .04) with annual faculty-departure rates for these departments (range = 2.5-19.6%). Fifty-six percent of the overall satisfaction scores fell in the unfavorable or lower half of the scale (1-5). The mean satisfaction scores ± 95% CI were strikingly similar between the departed and remaining groups (4.86 ± 0.51 versus 4.78 ± 0.34). Thirty-six percent of the respondents claimed to be on the tenure track. The overall satisfaction scores were not different between tenure- and non-tenure—track faculty (4.64 ± 2.20 versus 5.26 ± 2.29, p = .21).
Table 2 compares the differences in responses between faculty who had left the institution with those who remained. It also compares the differences in responses between those who gave overall satisfaction scores greater than and less than or equal to 5 (satisfied and unsatisfied, respectively). The greatest disparities between unsatisfied and satisfied respondents related to items of physician management: “trust—communication with chair/division head,” “equitable distribution of salary/resources,” “faculty involvement in department/division strategic planning,” and “written evaluation of your manager (e.g., section head).” In addition, logistic regression analyses showed that being non-tenure track, giving a low availability rating to “equitable distribution of salary/resources,” and giving a low availability rating to “positive collegial interaction” were found to be independent predictors of a low overall satisfaction rating. Analysis showed “equitable distribution of salary/resources” (p = .007), and “trust—communication with chair/division head” (p = .003) to be the independent variables in the logistic analysis with the other factors falling out of the model.
We analyzed and categorized the responses to the open-ended questions. When asked the reasons for continuing a career at the university, the primary categories of responses were lifestyle characteristics (nice community), 49%; intellectual characteristics, 46%; and humanitarian characteristics, 5%. The primary reasons physicians gave for leaving the university or for considering an opportunity elsewhere were administrative frustration, 59%; income enhancement, 18%; career advancement, 9%; academic frustration, 9%; and other reasons, 5%. Their recommendations for improving physicians' experiences at the university included fix administrative issues, 45%; improve research, 20%; increase income, 9%; more physician support, 9%; more program support, 8%; autonomy, 5%; and other suggestions, 4%.
The local community's lifestyle and intellectual gratification were the primary reasons the physicians gave for faculty-member retention. The lifestyle attraction was strong and balanced other, considerably negative, beliefs. A common theme was that the intensity of clinical work limited time for teaching and research. Resigning themselves to a “private practice” work environment (clinical care without teaching or research), faculty found that remuneration fell well short of that actually obtained in the private sector. Faculty members learned of salaries of comparable specialists through interaction with local colleagues in private practice, the Association of American Medical Colleges' faculty salary database, physician recruiters, and the occasional practice of the local newspaper to publish the salaries for each of the medical school's clinical faculty.
Many faculty members listed additional negative comments, with 72% listing an administrative concern serious enough to make them consider departing. Common among them were miscommunicated institutional priorities, insufficient financial support of clinical programs, poor billing and collection efforts, lack of control over out-patient clinic staff and policy, and inequalities between faculty members' on-call responsibilities and other clinical work. Income concerns usually were not cited directly, although many comments of administrative frustration were related to an enterprise failure to collect physician fees or reward work. Other concerns were directed at a physician—leader (e.g., department chair) if the relationship was strained; otherwise, hospital administrative personnel were blamed for problems with utilization of resources. There had been no formal exit interview process for most of the physicians who had left the medical center. Only 9% noted the potential for career advancement as a reason to consider an opportunity elsewhere.
This study shows that useful information regarding physician turnover can be derived from a relatively simple survey, which should be a routine exit procedure. Certain causes of physician turnover in an academic environment are beyond the effect of managerial intervention, such as stresses caused by commuting, community conditions, and external financing forces. Also, an academic institution should expect that its successful faculty members will advance their careers by finding leadership opportunities elsewhere. In our case, and we expect other academic health centers share our concerns, costly faculty turnover is relatively high despite favorable community conditions. We feel, therefore, that targeting different leadership practices that affect physicians' satisfaction would be useful.
Our study's findings confirm what other studies show, that good faculty—chairperson relations are associated with better work satisfaction, including a sense of professional achievement and satisfaction with academic autonomy, compensation, teaching, and research.6 Faculty in one survey wanted more time for research and scholarly work. Gender, race or ethnicity, academic rank, tenure status, and length of time at the institution also influenced faculty physicians' satisfaction.6,7 Favorable perceptions of global work satisfaction, autonomy in the workplace, professional status, teaching activities, clinical resources and activities, professional relationships, and institutional governance all correlate inversely with intentions of leaving.8
It is important to recognize that low physicians' satisfaction may not be immediately obvious to administrative leaders. McCullough, Dodge, and Moeller looked at factors that physicians deemed important in making a decision to affiliate with a health care organization.9 They examined both financial and non-financial incentives, yielding results similar to those already discussed. The importance physicians placed on the incentives was then compared with the importance placed on the incentives by administrators, and of 20 incentives examined, the groups differed significantly in their ratings of 12.
The insights gained from these data strongly support periodic surveys of physicians as a management tool. Faculty members within their first decade of institutional experience may have fewer support mechanisms, probably have concerns similar to those who have resigned, and comprise a population that may be amenable to successful interventions. Timely, comprehensive interventions—particularly for departments with high departure and low satisfaction rates—may be critical. Interventions may require better management practices, praise, or rewards for activities in all missions (research, education, and service). Programs that free time for faculty development, provide autonomy, and engage physicians in the process of managing scarce resources address many concerns cited in this study.
Within a traditional academic organizational structure, the faculty—chairperson relationship is crucial. With the growing complexities of academic medicine, rewarding the performances of many faculty members with diverse commitments to research, education, and service becomes difficult. Perhaps chairpersons should employ systems (comprehensive incentive plans, matrix management, etc.) that remain in place during the times of key leadership transitions. These systems can exist as formal arrangements where assigned managers focus on mission-specific tasks along service lines.10 Alternatively, rather than managing from the “top down,” some organizations define matrix management conceptually by influencing the attitudes and behaviors of first-line managers.11
Redundancy in physician leadership is important because some key physician leaders require months or years to replace. Not only are systems more robust than a single individual, they are also more resistant to perceptions of inequality that may occur if faculty—leadership relationships became strained.
Finally, an important method of limiting the perceived threat to physicians' autonomy and self-control caused by unstable or ineffective administration is to engage promising faculty in leadership training and other interdisciplinary professionalism-building activities. At the MUHC, the Millennium Health Care: Professional Development Course prepares physicians to assume leadership responsibilities in the academic setting and to overcome the inherent barriers to administration—faculty interaction (for complete details on the course, contact the authors). A total of 48 participants of the 1999 and 2000 Millennium classes were surveyed six months after completing the course. On a standard five-point scale their responses were very favorable or favorable (combined %) with respect to desire to assume new leadership roles (92%), confidence in leadership abilities (96%), ability to use coping strategies (92%), willingness to initiate change (92%), and understanding of challenges facing academic medical centers (98%). Similarly, they strongly agreed or agreed with items indicating that the other class members formed a network that could be called upon for support (88%). And they responded that they had formed new strategies for problem solving in their department (88%) as well as across departmental boundaries (92%).
While many of the concerns found in our surveys may be echoed elsewhere, this study is limited and not generalizable owing to its small sample size and single-institution focus. Because AHC's are complex, unique systems, it is likely that our survey would require tailoring to fit specific local concerns.
We conclude that concerns disclosed by survey data from relatively new faculty members at a single institution mirror those of departed faculty. Accordingly, these data can be used to improve faculty retention. It is likely that successful interventions will come about by significant alterations or enhancements of traditional academic physician leadership structures.