Academic and Career Development of Pulmonary and Critical Care Physician-Scientists
http://www.100md.com
《美国呼吸和危急护理医学》
Division of Pulmonary, Allergy, and Critical Care Medicine, and Clinical Outcomes Research Center, University of Minnesota Medical School, Minneapolis, Minnesota
Division of Neonatology, Department of Pediatrics, SUNY Buffalo and Women and Children's Hospital of Buffalo, Buffalo, New York
ABSTRACT
Rationale: The supply of research-oriented physicians is declining.
Objectives: Define barriers to successful career development in academic pulmonary/critical care medicine and discover strategies that promote the recruitment and retention of research physicians.
Methods: Focus groups and Web-based survey of fellows and junior faculty in pulmonary/critical care and neonatology divisions in 2002. Primary survey areas were educational debt, mentoring, institutional resources, information needs, and academic productivity.
Main Results: Monthly educational debt payments were a small fraction of junior faculty household income, but the debt/income ratio was larger for fellows and less for female faculty. The debt/income ratio was not associated with the perceived likelihood of an academic career. Respondents felt they lacked understanding of the roadmap to success and formal assistance in career development. Mentors were perceived as personally supportive, but less helpful in career development. Perceived likelihood of an academic career was most associated for faculty with institutional research support and for fellows with mentors' academic advising. Better research skills and more career development activities were associated with respondents having a Ph.D. mentor. Perceptions about academic physicians' job security were pessimistic.
Conclusions: Multiple factors influence decisions to pursue an academic medical career. In addition to alleviating financial pressures, academic careers may be aided by (1) providing more information about career pathways, job expectations, and success rates; (2) increasing career development mentoring; and (3) implementing formal curricula similar to Ph.D. programs. For pulmonary/critical care medicine faculty, current federal educational debt relief programs may have only a modest effect on academic retention.
Key Words: academic medicine biomedical research fellowships medical faculty mentors pulmonary
In many medical disciplines, the recruitment and retention of physician-scientists has decreased. A National Institutes of Health (NIH) workshop and several reviews of this problem suggest that the number of physician-scientists has fallen to a critically low level (1–3). A decline in research-oriented physicians likely would have major adverse effects on the research mission of academic medical centers and continued progress in medical science. Proposed interventions to reverse this trend include changing mentoring practices, greater visibility of physician- scientists to trainees, encouraging research experiences in medical school, increasing NIH K-series career development grants for junior faculty, and decreasing financial disincentives to pursue a research career. The latter suggestion led to the NIH Loan Repayment Program in 2001 (4).
To define barriers to successful career development and to discover strategies that promote recruitment and retention of physician in pulmonary/critical care research, we conducted a survey of fellowship trainees and junior faculty in adult and pediatric pulmonary medicine, critical care, and neonatology. The survey objectives were to (1) define the extent to which educational debt is a barrier to pursuing a physician-scientist career, (2) characterize the mentoring experience for trainees and junior faculty, (3) describe normative data on academic productivity benchmarks, and (4) identify perceived barriers to advancement and career satisfaction.
METHODS
We chose four primary content areas based on a literature review (1, 3, 5, 6) and analysis of focus groups held at the 2001 American Thoracic Society International Conference. The focus group methodology and analysis is in the online supplement.
Influence of Educational Debt
Educational debt was obtained as a monthly dollar analysis rather than total educational debt, as often is presented (5, 7, 8). We reasoned that many other personal cash-flow calculations are framed on a monthly household "cost versus income" basis and that a debt payment of $500/mo is of greater significance to a household with a monthly pretax income of $3,000 compared with a household with a $10,000 monthly income. We also reasoned that without access to detailed loan records, respondents would more likely remember their monthly debt payments compared with a loan principal balance or total loan cost (aggregate cost of monthly payments over the remaining life of the loan assuming no interest rate changes).
Respondents provided (within ordered ranges) their annual institutional pretax salary (plus bonuses if applicable, but excluding moonlighting income). We then assigned the dollar mean of the category range to each respondent. Respondents also chose a range category for the proportion of their salary to total household income, thereby permitting calculation of annual and monthly household income. The quotient of monthly educational debt payment and total monthly household income was the educational debt/income ratio. Because of the likely difficulty in estimating tax payments for respondents completing a 10-min e-mail survey, we did not adjust income for tax obligations. Because we wished to define debt burden as objectively as possible, the survey did not measure participants' perception or attitudes about the significance of the debt.
Additional details on the survey format, exact item wording and methods for evaluating mentorship quality, defining normative academic productivity, and perceived barriers to advancement are in the online supplement.
Study Sample
The study population was Postclinical Year 1 M.D. fellows in U.S. and Canadian training programs in adult or pediatric pulmonary, critical care, or combined programs, and neonatology as well as junior M.D. faculty (instructor or assistant professor rank) from the same divisions. Detailed information on defining the sampling frame, obtaining lists of e-mails, and defining survey methodology and response rates are in the online supplement (9). E-mails were sent to 1,374 distinct individuals, and the response rate was 46% (10).
The 95% confidence interval for 531 subjects giving a 50:50 response distribution to a dichotomous item (most conservative case) was 4.3%. Associations between ordered categoric variables were tested using Kendall's or Spearman's . Both analyses yield a statistic that ranges from –1 to +1; 2 x 2 tables were analyzed with Fisher's exact test. Grouped continuous variables were compared by the nonparametric Mann-Whitney test. Data were analyzed with SPSS version 11 (SPSS, Chicago, IL). The survey was approved by the University of Minnesota's institutional review board.
RESULTS
Responses were obtained from subjects residing in 44 states and Canada. Of the 531 respondents completing the form, 277 were fellows and 254 were self-defined as faculty (35 instructors, 199 assistant professors, 17 associate professors, and 3 "other"). We retained the responses of the 7% of faculty that were self-designated associate professors, assuming these were recent promotions.
Sex distribution, slightly favoring men, was similar in both ranks (Table 1). The mean faculty age of 38 was 5 yr greater than that of fellows, and the faculty had 5 more years of professional development since starting fellowship. The faculty were more likely to have children. The sample is dominated by trainees and faculty from adult pulmonary and critical care programs. The great majority of respondents provided some patient care during the 2001–2002 academic year. Respondents also had substantial research responsibilities: 90% of subjects in both the faculty and fellows group participated in one or more of the professional activity categories designated "research." Half the faculty reported spending 40% or more of their time on research and 56% of fellows were defined as research-intensive.
Effect of Educational Debt Burden on Physician-Scientist Career Choices
More than a quarter of fellows, but less than 3% of the faculty, were deferring some or all of their loans (Figure 1). Less than 10% of faculty had monthly educational debt payments of greater than $1,000/mo. In two-thirds of faculty households, there was another person contributing income. Based on the percent contribution of the faculty subjects' institutional income (excluding moonlighting income) to the total household income, the annual median faculty household income was $150,000 (interquartile range, $112,500–220,000). There was a weak inverse relationship between faculty educational debt payments and six ordered categories of total household income (Figure 2). Faculty had a low educational debt payment/household income ratio, with a median of 1.3% (interquartile range, 0.8–4.0%).
Not surprisingly, the fellows' median calculated household income of $56,000 (interquartile range, $45,000–93,000) was substantially less than that of faculty. This value does not include any moonlighting income. Fellows' debt/income ratio also was greater than the faculty ratio at 3.6% (interquartile range, 2.1–8.0%). There was a weak positive relationship between fellows' educational debt payments and total household income (this includes the 15% of fellows with debt payments or other loans in deferment; Figure 2). We cannot determine whether a fellow's decision to reduce debt payments by deferment was influenced by his or her household income.
There was no association between faculty sex and either institutional salary or monthly debt payments. However, the proportion of a faculty physician's salary contribution to their total household income was related to sex. Female faculty contributed a significantly lower percentage of total household income compared with male faculty ( = –0.31, p < .001), and 43% of men were the sole contributor to household income compared with only 20% of women. Because dual income was more common for women faculty, their total household income was greater than men (median $180,000 vs. $137,500, p < 0.01, Mann-Whitney test), which resulted in a slightly lower median debt/income ratio for women faculty (1.1%) compared with men (1.3%); difference significant at p < 0.001, Mann-Whitney test.
Respondents estimated their likelihood (low to high on a 1–5 scale) of being in an academic position in the next 3- to 4-yr period. Women respondents' estimated likelihood was lower than males, but the strength of the sex association was weak ( = 0.13, p = 0.02). An important question is whether the degree of debt influenced the likelihood of an academic career. For faculty, the distribution of perceived likelihood of remaining in academics was not related to monthly educational debt burden (Figure 3). For fellows, a similar lack of relationship was observed (Figure 4), but the overall distribution of perceived likelihood of remaining in academics was shifted to the left (i.e., lower overall likelihood compared with faculty). To account for the substantially different household incomes between the two groups, we plotted the five ordered intervals of the "academic likelihood" scale against the median debt/income ratio (Figure 5). For faculty, there was no relationship between perceived likelihood of staying in academics and median debt/income ratios. In addition, no likelihood group had a median debt/income ratio of greater than 1.4%. In contrast, fellows estimating a lower perceived likelihood of remaining in academics had higher median debt/income ratios, apart from the significant exception of the group most likely to stay in academics. Tempering this difference, this trend occurred within a relatively narrow range of median educational debt/income ratio categories and although fellows' debt/income ratios were greater than faculty ratios (because of fellows' lower incomes), the median value in all categories did not exceed 5%. The debt analysis for fellows is more complicated than for faculty because 26% of fellows were deferring some or all of their monthly debt payments. In summary, for fellows but not for current faculty, the level of debt may play a role in influencing the choice of an academic career.
For faculty, the perceived likelihood of remaining in academics was moderately associated with being defined as research-intensive ( = 0.30, p < .001), even though the debt/income ratios for research-intensive and nonintensive faculty were equal and low. Similar results were obtained for fellows, although the association was stronger ( = 0.53, p < 0.001).
These results suggest that, for current faculty, there is a relatively weak or no effect of educational debt on academic career plans. However, individuals whose career choice was debt sensitive may have already chosen not to enter a faculty position. Educational debt may be a stronger factor for fellows compared with faculty. Although faculty and fellows make career decisions based on numerous financial considerations including total salary, expected salary ladder, and noneducational debt, the independent influence of educational debt payments that do not rise above 5% of household income is likely to be small. This analysis is not based on the perceived effects of the debt, on the total household debt, or likely expenses from children or other sources, and the conclusions are tempered by the lack of this information.
Mentoring and Training
Table 2 lists the response formats for the three mentoring and research support scales, the items comprising the scale, the means and standard deviations and the coefficients. The two unmodified scales had an interitem correlation statistic comparable to the original report (6). The Institutional Research Support scale defined for this study by condensing the other two original scale components also had a similarly high value.
Within the Mentor Advising scale, both faculty and fellows rated highest their mentors' availability to critique their scientific work compared with their mentors' frequency of advising about their academic or career progress. Respondents' personal relationship with their mentors was very good, with fellows consistently giving slightly higher ratings of their mentors compared with junior faculty. In contrast, on the Institutional Research Support scale, faculty rated their overall research support as slightly worse relative to fellows. Research-intensive junior faculty gave more favorable summary scores on the Research support scale (lower scores = more favorable opinion) than non–research-intensive faculty (2.2 vs. 3.4, p < 0.001, Mann-Whitney test). The component with the worst rating by both groups was "availability of finances and time for research" with junior faculty rating the item more negatively than fellows.
Women faculty had equivalent scores to men in the Mentor Relationship and Institutional Research Support scale, but the ratings on the Mentor Advising scale (lower scores = worse opinion) were slightly lower (2.7 vs. 3, p = 0.017, Mann-Whitney test).
We next examined whether scores on the three scales were associated with physicians' plans for continuing in an academic position (the complete correlation matrix is in the online supplement). For both faculty and fellows, there were statistically significant correlations among the three scales and the perceived likelihood of future academics item. The Institutional Research Support scale rating was most strongly associated with plans for remaining in an academic position for junior faculty (r = –0.47), whereas the Mentor Relationship scale had the weakest association with academic likelihood in both groups (r = 0.23 and 0.16). The adequacy of mentor advising was more highly correlated with likelihood of an academic career for fellows than for faculty. For faculty, remaining in academics was most strongly associated with high ratings in research training quality and current research support (e.g., time, equipment, collaborators). Although these results show that the factors that influence physicians' plans for an academic career were different for fellows and faculty, in both groups the weakest correlate of academic career likelihood was the interpersonal quality of the mentoring relationship.
Fellows identified a mean of 4.7 persons whom they considered an "important" mentor to them during their fellowship. The survey did not ask respondents to qualify the mentors' role such as research or clinical mentor. A total of 62% of fellows were mentored by at least one individual holding a Ph.D. degree (Ph.D. or M.D./Ph.D.).
Faculty identified a mean of 4.6 persons whom they considered "important" mentors and 66% were mentored by at least one individual holding a Ph.D. degree. For both faculty and fellows, research-intensive physicians were more likely to have had a Ph.D. mentor (1.4 and 1.3 times more likely, respectively) compared with non–research-intensive physicians (p < 0.001, exact test). In both groups, having Ph.D. mentorship was weakly associated with self-rating of their research skills ( = 0.12, p = 0.04, and = 0.21, p < 0.001, respectively).
The quantity of formal career development activity was very low in both groups. More than half of all faculty had participated in less than 3 h of formal training since their hiring, whereas slightly more than a quarter had participated in more than 8 h of activity. Two-thirds of fellows had less than 3 h of development activities and about 20% had more than 8 h. There were no differences by sex. For both groups, having a Ph.D. mentor was weakly associated with participating in more hours of career development activities (faculty: = –0.14, p = 0.009, fellows: = –0.14, p = 0.012).
Productivity Norms
Table 3 shows the 25th, 50th, and 75th percentile values for two major academic productivity benchmarks: published papers and grant writing. Predictably, faculty published more papers and submitted or received funding for more grants than fellows. The publication-related questions asked respondents to start counting publications from the beginning of medical school, but the number of years (denominator) was defined as since the start of fellowship; therefore, people who published articles before fellowship would overestimate their "rate per year." Nevertheless, junior faculty members published, on average, 1.5 articles/yr; about half of those were research papers. Although the median number of publications per year by research-intensive faculty was statistically greater than by non–research-intensive faculty, the absolute difference was modest: 1.7 versus 1.1 (p < 0.001, Mann-Whitney). Even at the 75th percentile, junior faculty published at a rate of less than one article per year as a major contributor (defined as publishing first, second, or last authored research papers). The median number of total grant applications submitted for faculty was three, with most grant applications submitted to national organizations.
Informational Needs and Perceived Barriers to Advancement
Respondents' greatest need for information concerned issues of hiring and promotion (Figures 6 and 7), whereas developing mentoring relationships was in the lower half of importance. Faculty were most concerned about "methods to increase their visibility as productive faculty members," with significant concerns about achieving promotion and balancing clinical work and research. In contrast, more than two-thirds of fellows desired to receive more advice on transitioning to a new junior faculty position. Approximately one-third of fellows wanted more information on clinical practice options, but this was important for only 1 of 10 faculty members.
Two-thirds of faculty strongly agreed that research productivity and obtaining grant support were important, whereas less than 20% strongly agreed that clinical teaching and patient care activities were important for their careers (complete data are available in the online supplement). Fellows rated clinical teaching and patient care activities as more important than faculty. Fellow rating of the importance of research productivity was strongly related to their perceived likelihood of being in an academic position in the near future ( = 0.6, p < 0.001).
The focus groups (online supplement) suggested that uncertainty of academic career success was a major concern for junior trainees. In response to the statement "relative to the practice of clinical medicine, a physician-scientist career offered job security," only 4% of faculty indicated strong agreement, 9% indicated agreement of lesser intensity, and 29% strongly disagreed. Neither faculty research-intensive status nor sex was associated with the degree of agreement. Of fellows, 7% strongly agreed with the statement that a physician-scientist career offered job security, whereas 13% indicated agreement of lesser intensity and 25% strongly disagreed.
DISCUSSION
This is the largest subspecialty-specific survey of academic career development in medicine that we have identified. Results from this survey describe the similarities and differences between fellowship trainees and the junior faculty who have accumulated approximately 5 more years of career experience. The survey and the focus groups identified multiple components that affect the likelihood of pursuing a career in academic pulmonary and critical care medicine: formal training in career development, the research and career advising components within mentoring relationships, involvement with Ph.D. mentors, institutional support for research, time available for research, lack of knowledge about academic jobs and milestones for successful career advancement, and uncertainty about academic career success. For fellows, debt may play a role in career choice, but there is not a very strong association supported by our data.
The faculty group had a relatively low debt:income ratio of 1.3%, and an approximate median household income of $150,000/yr, aided by the fact that two-thirds of the faculty households had other individuals contributing income. Only 20% of the female faculty, but 43% of the male faculty, were the sole household income generators. Female and male faculty did not differ in salary or debt levels. The self-rated likelihood of staying in academics was not associated with debt level, but did associate with the faculty member's degree of interest in research. Faculty identified a relatively large number of mentors (4.6) and were positive about personal aspects of mentoring, but rated the degree of institutional research support and critical mentoring of their work as less well supported. The quality of mentoring was associated with the perceived likelihood of staying in academics. Faculty had received surprisingly little formal education related to research and career development (e.g., grant writing), with more than 50% of faculty having 3 h or less of formal instruction and only 25% with at least 8 h. The publishing productivity of junior faculty was surprisingly low, with a mean of 1.5 manuscripts published per year, and approximately half of these were original research. The faculty served as a major contributor (first, second, or last author) on less than one manuscript per year. The junior faculty principally desired additional information about hiring and promotion policies and about how to increase their professional visibility.
The fellows group had many similarities with the faculty, but some differences. Approximately one-quarter of the fellows were deferring their debt repayment and the remainder had more debt than the faculty, with a median debt:income ratio (not including those in deferral) of 3.6%. The approximate median annual household income, excluding moonlighting, was $56,000. In contrast to the faculty, there was an association of lower academic likelihood with higher debt:income ratio, except for a group that was highly dedicated to an academic career—apparently in spite of their debt. Fellows also identified a large number of mentors (mean, 4.7), but in contrast to the faculty, they felt that mentor advising was the most important function. The fellows also received little formal instruction related to career development, with two-thirds receiving 3 or fewer hours of instruction. The fellows' major desire was for more information about transitioning their careers to faculty status (or practice). They rated clinical care and teaching as more important functions than did the faculty.
One surprising primary finding is that for junior faculty the amount of educational debt was not associated with future career plans when analyzed by a monthly debt/income method. This conclusion is supported by demonstrating that (1) the proportion of faculty with very high monthly payments is low, (2) faculty salaries are usually greater than $100,000/yr, (3) many faculty are in two-income families with the other person often in a high-salary position, (4) there was no association between amount of monthly educational debt payments and estimated likelihood of continuing in an academic career, and (5) research-oriented faculty and fellows had equivalent debt/income ratios as their counterparts without significant protected research time or interests. Although there undoubtedly are "outlier" scenarios of faculty with high educational debt payments and low institutional salaries who have non–wage-earning spouses and children living in an expensive metropolitan area, our data demonstrate that, for most faculty, educational debt payments are less than 5% of total household pretax income. In addition, women faculty are more likely to be in a two-income household and have lower educational debt pressures. These results are significant because they raise the possibility that current federal interventions to reduce educational debt will substantially affect the finances of only a small proportion of faculty who have salaries similar to physicians in this survey. We predict that, by itself, the NIH Loan Repayment Program will have only an overall modest effect on recruiting and retaining physician-scientists.
The faculty data may be biased in that the survey did not sample physicians that, based on their debt burden or pessimistic financial calculations during fellowship, never considered a faculty position and entered private practice before the date of the survey. However, the fellows' data (Figure 4) also shows no strong relationship between educational debt payment and academic career plans. In contrast, Figure 5 shows a relationship between lower debt/income ratio and greater likelihood of academic career for fellows (with the exception of the last category), but also shows that the range of the median monthly debt/income ratio across all the categories of academic likelihood is only 1.5%. We acknowledge that defining the finances of individuals is complex, and this survey was not designed to account for tax obligations, noneducational debt, spousal debt, or lifestyle expectations. On the other hand, because we asked respondents to calculate only their institutional income, it is likely that some fellows and faculty supplement their income with noninstitutional employment (moonlighting or consulting). For inpatient-based specialties, such as pulmonary and critical care medicine, noninstitutional income can be substantial. We did not specifically determine the total educational or individual debt of individual respondents or their households; future surveys could compare these different measures of debt and their association with career decision making. Career choices also may also be influenced by physicians' expectations of future income, which we did not estimate.
The fiscal situation of fellows is different from junior faculty. Fellows' overall debt/income ratio was almost three times higher than the faculty ratio and financial pressures likely account for the one-quarter of fellows who are in full or partial deferment of loan payments. However, this tight financial situation of the fellows is likely to be short-lived, because we show that taking a faculty position in a pulmonary or critical care division greatly reduces educational debt pressures and thereby may lessen the influence of debt on continuing in academics. It is uncertain how much fellows know of the financial and debt conditions of typical junior faculty members. It is uncertain whether the NIH Loan Repayment Program, which solely offers repayment of educational debt principal (i.e., does not immediately reduce monthly payments), has a large financial impact on attracting and retaining junior research faculty in pulmonary/critical care positions. There may be, however, other nonfinancial benefits to recipients of a Loan Repayment Award, such as professional recognition. Disseminating detailed financial data such as salary distributions, pay ladders, and average indebtedness after 5 or more years of faculty status may positively influence fellows' decisions to consider an academic career by alleviating concern over the degree of persistent indebtedness for junior faculty.
It is possible that some respondents in the "faculty" group were actually fellowship trainees who had been given the rank of instructor, as is the custom at some programs. This is unlikely to affect the overall faculty results substantially as instructors comprised only 14% of the faculty group. The 5-yr mean age difference between the faculty and fellow groups suggests that a large number of "older fellows" were not misclassified as "young faculty."
The mentor–protege relationship is multifaceted, and our results show that a relatively large number of persons (mean, 4.6) are perceived as "important" mentors from fellowship through assistant professorship. Mentoring for women faculty was similar to men: the major exception was the Mentor Advising scale, which assessed items related to promoting the protege's visibility and advice about academic progress. Both fellows and faculty rated highly their mentors' personal relationship with them, but mentors were more likely to advise trainees on their scientific work rather than on more global topics, such as academic or career progress. This discrepancy may be particularly important to junior faculty because the most highly desired information categories are academic issues of promotion, visibility, and balancing competing responsibilities. Mentors may not consider these areas of information relevant to the mentor–protege relationship, or these areas may be inherently more ambiguous than teaching the scientific method and reviewing data. Alternatively, mentors' own time and productivity pressures may place practical limits on the extent of discussion. Similarly, for fellows, the major area of hard-to-get information was how to advance to the next academic level. These results and comments from the focus groups suggest that developing physician-scientists very much desire and would benefit from an "academic roadmap" that addresses these topics.
Two analyses suggest that mentors' own academic training history influences the quality and preparation of their trainees. Both faculty and fellows who had mentors holding Ph.D. degrees had higher self-reported research skills and participated in more organized research or career development activities. These associations were weak, but consistent. The reason for this association cannot be determined from this survey design, but a logical explanation is that a mentor's graduate school experience (e.g., graded classes, examinations and defenses, thesis advisory committees, thesis writing) creates training standards that the mentor expects their M.D. trainees to meet. Undoubtedly, there are excellent M.D. mentors without Ph.D. degrees, but our findings suggest that incorporating specific graduate school training concepts into fellowship programs could increase the number of successful physician-scientists (11). This conclusion is somewhat tempered by recent reports that university-based research Ph.D. programs in both the sciences and the humanities have experienced development and retention problems similar to that of physician-scientists (12, 13). It is also possible that Ph.D. mentorship has no beneficial independent effects on the scientific skills of faculty or fellows, but that the association is confounded by highly motivated physicians' self-selection to train at institutions with strong graduate programs.
Annual reviews and meetings with supervisors have long served as formal mechanisms to assess academic progress, but peer normative data can also serve as an independent marker of achievement and career advancement. Although different institutions have varying expectations of research productivity as measured in publications or grant support, this survey suggests that less than one primary-author research publication per year represents average progress. About half of all publications by junior faculty are considered nonresearch publications. This fact should lead to further dialog among junior and senior faculty regarding the weight of nonresearch publications in academic career development and preparation for promotion decisions.
Job security was an issue repeatedly raised in the focus groups. It is logical that trainees who perceive their odds of sustaining a physician-scientist career as low would be less likely to commit their time and opportunity costs to train for that career. For instance, participants believed that their specialized research skills would be useless in a community practice job if they could not remain in academic medicine. Our finding that only 13% of faculty and 20% of fellows agreed that an academic career had more job security than a clinical practice position may explain much of the difficulties in attracting and retaining physicians in an academic career. Concerns about academic job security seem to have not abated since the survey of pulmonary and critical care junior faculty survey by Broaddus and Feigal in 1988 (14). An academic research career requires competing on a national level and even committed, well trained persons may be unsuccessful (15). However, there are few data available to trainees that allow them to make an accurate appraisal of the career success and retention rates for peer physician-scientists. Collecting and disseminating such data may offset some unduly pessimistic attitudes about the viability of an academic career in medicine. It would be very helpful for professional societies to develop methods to prospectively track the careers of a cohort of trainees and junior faculty. Similarly, tracking of the success rates and careers of junior grant awardees from the NIH, American Heart Association, and American Lung Association would be useful information for advisors, fellows, and junior faculty. Whether the solution to the issue of career uncertainty is to try to reduce it for faculty or to insulate trainees from this faculty concern is uncertain.
A limitation of this report is that the survey was focused on pulmonary and critical care–oriented physicians with respondents mostly practicing adult medicine. Although the survey touched on issues likely relevant to all trainees and junior faculty, it is unknown whether the specific results are generalizable to other subspecialties within pediatrics or internal medicine or to specialties such as surgery or psychiatry. Other variables that were not examined in this survey (the presence of children; spousal career nature; and personal or spousal expectations about financial stability) also may play roles in career decision making and would be important areas for future investigations. A response rate of 46% may create bias in the estimates of group characteristics, such as average age or monthly debt payments, but response bias should have a lesser effect on correlational analyses within the survey such as the relationship between debt payments and future career plans.
In summary, a trainee's decision to prepare and continue in a physician-scientist career has multiple determinants and incorporates personal characteristics, academic environment, work, and domestic lifestyle expectations, financial reward, and job security. If the future productivity of the nation's biomedical workforce and the mission of academic health centers is at least partially dependent on developing and retaining academic physician- scientist, then, based on our findings, we conclude that amelioration of educational debt is only one aspect of an effective solution. For fellows interested in an academic career, and for faculty, our data suggest that effective interventions to improve physician-scientist training and retention would include greater discussion of the academic career ladder and job security by mentors, formal career development meetings (supplemented by empiric data about productivity and timelines to grant success), processes to increase professional visibility, and implementing educational methods common to doctoral research programs. Identifying ways to reduce the anxiety and concern about career success in research for individuals who are committed to this pathway may also be an important method to alleviate this long-term problem. It is not known whether these interventions would be most effectively coordinated by divisional, departmental, school, or professional society organizations.
Acknowledgments
The authors appreciate the assistance of Todd Rockwood, Ph.D., in revising the survey format and content; Monika Simeonova and Elisha Malanga of the American Thoracic Society, Darlene Buczak of the Association of Pulmonary and Critical Care Program Directors and the Chest Foundation for assistance in compiling the survey list; Joan Watkins for transcription of the focus groups; and Courtney Broaddus, M.D., of University of California–San Francisco and Peter Bitterman, M.D., at the University of Minnesota for advice on survey content and analysis.
FOOTNOTES
Supported by a grant from the American Thoracic Society.
Versions of this research were published in abstract form in Am J Respir Crit Care Med 2002;165:A110.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.200503-325OC on September 22, 2005
Conflict of Interest Statement: C.R.W. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. J.B. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. R.R. has participated as a speaker in a conference that was partially financed by Ross, although the lectures were unrelated to their products. She owns stock in Abbot Laboratories as part of a portfolio. D.H.I. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
REFERENCES
National Heart, Lung and Blood Institute. Research training and career development: human resources for research in the 21st century: report of a November 1999 workshop. Bethesda, M.D.: National Heart, Lung, and Blood Institute. Available from: http://www.nhlbi.nih.gov/funding/training/train-rev.pdf (accessed November 12, 2004).
Lenfant C. Training the next generation of biomedical researchers: challenges and opportunities. Circulation 2000;102:368.
Rosenberg L. The physician-scientist: an essential—and fragile—link in the medical research chain. J Clin Invest 1999;103:1621–1626.
Ley T, Rosenberg L. Removing career obstacles for young physician-scientists: loan repayment programs. N Engl J Med 2002;346:368–372.
Zemlo T, Garrison H, Partridge N, Ley T. The physician-scientist: career issues and challenges at the year 2000. FASEB J 2000;14:221–230.
Palepu A, Friedman R, Barnett R, Carr P, Ash A, Szalacha L, Moskowitz M. Junior faculty members' mentoring relationships and their professional development in US medical schools. Acad Med 1998;73:318–323.
Kazzi A, Langdorf M, Brillman J, Handly N, Munden S. Emergency medicine residency applicant educational debt: relationship with attitude toward training and moonlighting. Acad Emerg Med 2000;7: 1399–1407.
Colquitt W, Zeh M, Killian C, Cultice J. Effect of debt on US medical school graduates' preferences for family medicine, general internal medicine and general pediatrics. Acad Med 1996;71:399–411.
National Institutes of Health awards to medical schools (medicine departments) fiscal year 2001. Available from: http://grants.nih.gov/grants/award/rank/medicine01.htm (accessed November 26, 2004).
The American Association for Public Opinion Research. Standard definitions: final dispositions of case codes and outcomes rates for surveys. Ann Arbor, MI: AAPOR; 2000.
Varki A, Rosenberg L. Emerging opportunities and career paths for the young physician-scientist. Nat Med 2002;8:437–439.
Golde C, Dore T. At cross purposes: what the experiences of doctoral students reveal about doctoral education. Report prepared for The Pew Charitable Trusts, Philadelphia, 2001. Available from: http://www.phd-survey.org (accessed November 12, 2004).
Council of Graduate Schools. Ph.D. completion and attrition: policy, numbers, leadership, and next steps. Washington, DC: Council of Graduate Schools; 2004.
Broaddus V, Feigal D. Starting an academic career: a survey of junior academic pulmonary physicians. Chest 1994;105:1858–1863.
Kotchen T, Lindquist T, Malik K, Ehrenfield E. NIH peer review of grant applications for clinical research. JAMA 2004;291:836–843.(Craig R. Weinert, Joanne Billings, Rita )
Division of Neonatology, Department of Pediatrics, SUNY Buffalo and Women and Children's Hospital of Buffalo, Buffalo, New York
ABSTRACT
Rationale: The supply of research-oriented physicians is declining.
Objectives: Define barriers to successful career development in academic pulmonary/critical care medicine and discover strategies that promote the recruitment and retention of research physicians.
Methods: Focus groups and Web-based survey of fellows and junior faculty in pulmonary/critical care and neonatology divisions in 2002. Primary survey areas were educational debt, mentoring, institutional resources, information needs, and academic productivity.
Main Results: Monthly educational debt payments were a small fraction of junior faculty household income, but the debt/income ratio was larger for fellows and less for female faculty. The debt/income ratio was not associated with the perceived likelihood of an academic career. Respondents felt they lacked understanding of the roadmap to success and formal assistance in career development. Mentors were perceived as personally supportive, but less helpful in career development. Perceived likelihood of an academic career was most associated for faculty with institutional research support and for fellows with mentors' academic advising. Better research skills and more career development activities were associated with respondents having a Ph.D. mentor. Perceptions about academic physicians' job security were pessimistic.
Conclusions: Multiple factors influence decisions to pursue an academic medical career. In addition to alleviating financial pressures, academic careers may be aided by (1) providing more information about career pathways, job expectations, and success rates; (2) increasing career development mentoring; and (3) implementing formal curricula similar to Ph.D. programs. For pulmonary/critical care medicine faculty, current federal educational debt relief programs may have only a modest effect on academic retention.
Key Words: academic medicine biomedical research fellowships medical faculty mentors pulmonary
In many medical disciplines, the recruitment and retention of physician-scientists has decreased. A National Institutes of Health (NIH) workshop and several reviews of this problem suggest that the number of physician-scientists has fallen to a critically low level (1–3). A decline in research-oriented physicians likely would have major adverse effects on the research mission of academic medical centers and continued progress in medical science. Proposed interventions to reverse this trend include changing mentoring practices, greater visibility of physician- scientists to trainees, encouraging research experiences in medical school, increasing NIH K-series career development grants for junior faculty, and decreasing financial disincentives to pursue a research career. The latter suggestion led to the NIH Loan Repayment Program in 2001 (4).
To define barriers to successful career development and to discover strategies that promote recruitment and retention of physician in pulmonary/critical care research, we conducted a survey of fellowship trainees and junior faculty in adult and pediatric pulmonary medicine, critical care, and neonatology. The survey objectives were to (1) define the extent to which educational debt is a barrier to pursuing a physician-scientist career, (2) characterize the mentoring experience for trainees and junior faculty, (3) describe normative data on academic productivity benchmarks, and (4) identify perceived barriers to advancement and career satisfaction.
METHODS
We chose four primary content areas based on a literature review (1, 3, 5, 6) and analysis of focus groups held at the 2001 American Thoracic Society International Conference. The focus group methodology and analysis is in the online supplement.
Influence of Educational Debt
Educational debt was obtained as a monthly dollar analysis rather than total educational debt, as often is presented (5, 7, 8). We reasoned that many other personal cash-flow calculations are framed on a monthly household "cost versus income" basis and that a debt payment of $500/mo is of greater significance to a household with a monthly pretax income of $3,000 compared with a household with a $10,000 monthly income. We also reasoned that without access to detailed loan records, respondents would more likely remember their monthly debt payments compared with a loan principal balance or total loan cost (aggregate cost of monthly payments over the remaining life of the loan assuming no interest rate changes).
Respondents provided (within ordered ranges) their annual institutional pretax salary (plus bonuses if applicable, but excluding moonlighting income). We then assigned the dollar mean of the category range to each respondent. Respondents also chose a range category for the proportion of their salary to total household income, thereby permitting calculation of annual and monthly household income. The quotient of monthly educational debt payment and total monthly household income was the educational debt/income ratio. Because of the likely difficulty in estimating tax payments for respondents completing a 10-min e-mail survey, we did not adjust income for tax obligations. Because we wished to define debt burden as objectively as possible, the survey did not measure participants' perception or attitudes about the significance of the debt.
Additional details on the survey format, exact item wording and methods for evaluating mentorship quality, defining normative academic productivity, and perceived barriers to advancement are in the online supplement.
Study Sample
The study population was Postclinical Year 1 M.D. fellows in U.S. and Canadian training programs in adult or pediatric pulmonary, critical care, or combined programs, and neonatology as well as junior M.D. faculty (instructor or assistant professor rank) from the same divisions. Detailed information on defining the sampling frame, obtaining lists of e-mails, and defining survey methodology and response rates are in the online supplement (9). E-mails were sent to 1,374 distinct individuals, and the response rate was 46% (10).
The 95% confidence interval for 531 subjects giving a 50:50 response distribution to a dichotomous item (most conservative case) was 4.3%. Associations between ordered categoric variables were tested using Kendall's or Spearman's . Both analyses yield a statistic that ranges from –1 to +1; 2 x 2 tables were analyzed with Fisher's exact test. Grouped continuous variables were compared by the nonparametric Mann-Whitney test. Data were analyzed with SPSS version 11 (SPSS, Chicago, IL). The survey was approved by the University of Minnesota's institutional review board.
RESULTS
Responses were obtained from subjects residing in 44 states and Canada. Of the 531 respondents completing the form, 277 were fellows and 254 were self-defined as faculty (35 instructors, 199 assistant professors, 17 associate professors, and 3 "other"). We retained the responses of the 7% of faculty that were self-designated associate professors, assuming these were recent promotions.
Sex distribution, slightly favoring men, was similar in both ranks (Table 1). The mean faculty age of 38 was 5 yr greater than that of fellows, and the faculty had 5 more years of professional development since starting fellowship. The faculty were more likely to have children. The sample is dominated by trainees and faculty from adult pulmonary and critical care programs. The great majority of respondents provided some patient care during the 2001–2002 academic year. Respondents also had substantial research responsibilities: 90% of subjects in both the faculty and fellows group participated in one or more of the professional activity categories designated "research." Half the faculty reported spending 40% or more of their time on research and 56% of fellows were defined as research-intensive.
Effect of Educational Debt Burden on Physician-Scientist Career Choices
More than a quarter of fellows, but less than 3% of the faculty, were deferring some or all of their loans (Figure 1). Less than 10% of faculty had monthly educational debt payments of greater than $1,000/mo. In two-thirds of faculty households, there was another person contributing income. Based on the percent contribution of the faculty subjects' institutional income (excluding moonlighting income) to the total household income, the annual median faculty household income was $150,000 (interquartile range, $112,500–220,000). There was a weak inverse relationship between faculty educational debt payments and six ordered categories of total household income (Figure 2). Faculty had a low educational debt payment/household income ratio, with a median of 1.3% (interquartile range, 0.8–4.0%).
Not surprisingly, the fellows' median calculated household income of $56,000 (interquartile range, $45,000–93,000) was substantially less than that of faculty. This value does not include any moonlighting income. Fellows' debt/income ratio also was greater than the faculty ratio at 3.6% (interquartile range, 2.1–8.0%). There was a weak positive relationship between fellows' educational debt payments and total household income (this includes the 15% of fellows with debt payments or other loans in deferment; Figure 2). We cannot determine whether a fellow's decision to reduce debt payments by deferment was influenced by his or her household income.
There was no association between faculty sex and either institutional salary or monthly debt payments. However, the proportion of a faculty physician's salary contribution to their total household income was related to sex. Female faculty contributed a significantly lower percentage of total household income compared with male faculty ( = –0.31, p < .001), and 43% of men were the sole contributor to household income compared with only 20% of women. Because dual income was more common for women faculty, their total household income was greater than men (median $180,000 vs. $137,500, p < 0.01, Mann-Whitney test), which resulted in a slightly lower median debt/income ratio for women faculty (1.1%) compared with men (1.3%); difference significant at p < 0.001, Mann-Whitney test.
Respondents estimated their likelihood (low to high on a 1–5 scale) of being in an academic position in the next 3- to 4-yr period. Women respondents' estimated likelihood was lower than males, but the strength of the sex association was weak ( = 0.13, p = 0.02). An important question is whether the degree of debt influenced the likelihood of an academic career. For faculty, the distribution of perceived likelihood of remaining in academics was not related to monthly educational debt burden (Figure 3). For fellows, a similar lack of relationship was observed (Figure 4), but the overall distribution of perceived likelihood of remaining in academics was shifted to the left (i.e., lower overall likelihood compared with faculty). To account for the substantially different household incomes between the two groups, we plotted the five ordered intervals of the "academic likelihood" scale against the median debt/income ratio (Figure 5). For faculty, there was no relationship between perceived likelihood of staying in academics and median debt/income ratios. In addition, no likelihood group had a median debt/income ratio of greater than 1.4%. In contrast, fellows estimating a lower perceived likelihood of remaining in academics had higher median debt/income ratios, apart from the significant exception of the group most likely to stay in academics. Tempering this difference, this trend occurred within a relatively narrow range of median educational debt/income ratio categories and although fellows' debt/income ratios were greater than faculty ratios (because of fellows' lower incomes), the median value in all categories did not exceed 5%. The debt analysis for fellows is more complicated than for faculty because 26% of fellows were deferring some or all of their monthly debt payments. In summary, for fellows but not for current faculty, the level of debt may play a role in influencing the choice of an academic career.
For faculty, the perceived likelihood of remaining in academics was moderately associated with being defined as research-intensive ( = 0.30, p < .001), even though the debt/income ratios for research-intensive and nonintensive faculty were equal and low. Similar results were obtained for fellows, although the association was stronger ( = 0.53, p < 0.001).
These results suggest that, for current faculty, there is a relatively weak or no effect of educational debt on academic career plans. However, individuals whose career choice was debt sensitive may have already chosen not to enter a faculty position. Educational debt may be a stronger factor for fellows compared with faculty. Although faculty and fellows make career decisions based on numerous financial considerations including total salary, expected salary ladder, and noneducational debt, the independent influence of educational debt payments that do not rise above 5% of household income is likely to be small. This analysis is not based on the perceived effects of the debt, on the total household debt, or likely expenses from children or other sources, and the conclusions are tempered by the lack of this information.
Mentoring and Training
Table 2 lists the response formats for the three mentoring and research support scales, the items comprising the scale, the means and standard deviations and the coefficients. The two unmodified scales had an interitem correlation statistic comparable to the original report (6). The Institutional Research Support scale defined for this study by condensing the other two original scale components also had a similarly high value.
Within the Mentor Advising scale, both faculty and fellows rated highest their mentors' availability to critique their scientific work compared with their mentors' frequency of advising about their academic or career progress. Respondents' personal relationship with their mentors was very good, with fellows consistently giving slightly higher ratings of their mentors compared with junior faculty. In contrast, on the Institutional Research Support scale, faculty rated their overall research support as slightly worse relative to fellows. Research-intensive junior faculty gave more favorable summary scores on the Research support scale (lower scores = more favorable opinion) than non–research-intensive faculty (2.2 vs. 3.4, p < 0.001, Mann-Whitney test). The component with the worst rating by both groups was "availability of finances and time for research" with junior faculty rating the item more negatively than fellows.
Women faculty had equivalent scores to men in the Mentor Relationship and Institutional Research Support scale, but the ratings on the Mentor Advising scale (lower scores = worse opinion) were slightly lower (2.7 vs. 3, p = 0.017, Mann-Whitney test).
We next examined whether scores on the three scales were associated with physicians' plans for continuing in an academic position (the complete correlation matrix is in the online supplement). For both faculty and fellows, there were statistically significant correlations among the three scales and the perceived likelihood of future academics item. The Institutional Research Support scale rating was most strongly associated with plans for remaining in an academic position for junior faculty (r = –0.47), whereas the Mentor Relationship scale had the weakest association with academic likelihood in both groups (r = 0.23 and 0.16). The adequacy of mentor advising was more highly correlated with likelihood of an academic career for fellows than for faculty. For faculty, remaining in academics was most strongly associated with high ratings in research training quality and current research support (e.g., time, equipment, collaborators). Although these results show that the factors that influence physicians' plans for an academic career were different for fellows and faculty, in both groups the weakest correlate of academic career likelihood was the interpersonal quality of the mentoring relationship.
Fellows identified a mean of 4.7 persons whom they considered an "important" mentor to them during their fellowship. The survey did not ask respondents to qualify the mentors' role such as research or clinical mentor. A total of 62% of fellows were mentored by at least one individual holding a Ph.D. degree (Ph.D. or M.D./Ph.D.).
Faculty identified a mean of 4.6 persons whom they considered "important" mentors and 66% were mentored by at least one individual holding a Ph.D. degree. For both faculty and fellows, research-intensive physicians were more likely to have had a Ph.D. mentor (1.4 and 1.3 times more likely, respectively) compared with non–research-intensive physicians (p < 0.001, exact test). In both groups, having Ph.D. mentorship was weakly associated with self-rating of their research skills ( = 0.12, p = 0.04, and = 0.21, p < 0.001, respectively).
The quantity of formal career development activity was very low in both groups. More than half of all faculty had participated in less than 3 h of formal training since their hiring, whereas slightly more than a quarter had participated in more than 8 h of activity. Two-thirds of fellows had less than 3 h of development activities and about 20% had more than 8 h. There were no differences by sex. For both groups, having a Ph.D. mentor was weakly associated with participating in more hours of career development activities (faculty: = –0.14, p = 0.009, fellows: = –0.14, p = 0.012).
Productivity Norms
Table 3 shows the 25th, 50th, and 75th percentile values for two major academic productivity benchmarks: published papers and grant writing. Predictably, faculty published more papers and submitted or received funding for more grants than fellows. The publication-related questions asked respondents to start counting publications from the beginning of medical school, but the number of years (denominator) was defined as since the start of fellowship; therefore, people who published articles before fellowship would overestimate their "rate per year." Nevertheless, junior faculty members published, on average, 1.5 articles/yr; about half of those were research papers. Although the median number of publications per year by research-intensive faculty was statistically greater than by non–research-intensive faculty, the absolute difference was modest: 1.7 versus 1.1 (p < 0.001, Mann-Whitney). Even at the 75th percentile, junior faculty published at a rate of less than one article per year as a major contributor (defined as publishing first, second, or last authored research papers). The median number of total grant applications submitted for faculty was three, with most grant applications submitted to national organizations.
Informational Needs and Perceived Barriers to Advancement
Respondents' greatest need for information concerned issues of hiring and promotion (Figures 6 and 7), whereas developing mentoring relationships was in the lower half of importance. Faculty were most concerned about "methods to increase their visibility as productive faculty members," with significant concerns about achieving promotion and balancing clinical work and research. In contrast, more than two-thirds of fellows desired to receive more advice on transitioning to a new junior faculty position. Approximately one-third of fellows wanted more information on clinical practice options, but this was important for only 1 of 10 faculty members.
Two-thirds of faculty strongly agreed that research productivity and obtaining grant support were important, whereas less than 20% strongly agreed that clinical teaching and patient care activities were important for their careers (complete data are available in the online supplement). Fellows rated clinical teaching and patient care activities as more important than faculty. Fellow rating of the importance of research productivity was strongly related to their perceived likelihood of being in an academic position in the near future ( = 0.6, p < 0.001).
The focus groups (online supplement) suggested that uncertainty of academic career success was a major concern for junior trainees. In response to the statement "relative to the practice of clinical medicine, a physician-scientist career offered job security," only 4% of faculty indicated strong agreement, 9% indicated agreement of lesser intensity, and 29% strongly disagreed. Neither faculty research-intensive status nor sex was associated with the degree of agreement. Of fellows, 7% strongly agreed with the statement that a physician-scientist career offered job security, whereas 13% indicated agreement of lesser intensity and 25% strongly disagreed.
DISCUSSION
This is the largest subspecialty-specific survey of academic career development in medicine that we have identified. Results from this survey describe the similarities and differences between fellowship trainees and the junior faculty who have accumulated approximately 5 more years of career experience. The survey and the focus groups identified multiple components that affect the likelihood of pursuing a career in academic pulmonary and critical care medicine: formal training in career development, the research and career advising components within mentoring relationships, involvement with Ph.D. mentors, institutional support for research, time available for research, lack of knowledge about academic jobs and milestones for successful career advancement, and uncertainty about academic career success. For fellows, debt may play a role in career choice, but there is not a very strong association supported by our data.
The faculty group had a relatively low debt:income ratio of 1.3%, and an approximate median household income of $150,000/yr, aided by the fact that two-thirds of the faculty households had other individuals contributing income. Only 20% of the female faculty, but 43% of the male faculty, were the sole household income generators. Female and male faculty did not differ in salary or debt levels. The self-rated likelihood of staying in academics was not associated with debt level, but did associate with the faculty member's degree of interest in research. Faculty identified a relatively large number of mentors (4.6) and were positive about personal aspects of mentoring, but rated the degree of institutional research support and critical mentoring of their work as less well supported. The quality of mentoring was associated with the perceived likelihood of staying in academics. Faculty had received surprisingly little formal education related to research and career development (e.g., grant writing), with more than 50% of faculty having 3 h or less of formal instruction and only 25% with at least 8 h. The publishing productivity of junior faculty was surprisingly low, with a mean of 1.5 manuscripts published per year, and approximately half of these were original research. The faculty served as a major contributor (first, second, or last author) on less than one manuscript per year. The junior faculty principally desired additional information about hiring and promotion policies and about how to increase their professional visibility.
The fellows group had many similarities with the faculty, but some differences. Approximately one-quarter of the fellows were deferring their debt repayment and the remainder had more debt than the faculty, with a median debt:income ratio (not including those in deferral) of 3.6%. The approximate median annual household income, excluding moonlighting, was $56,000. In contrast to the faculty, there was an association of lower academic likelihood with higher debt:income ratio, except for a group that was highly dedicated to an academic career—apparently in spite of their debt. Fellows also identified a large number of mentors (mean, 4.7), but in contrast to the faculty, they felt that mentor advising was the most important function. The fellows also received little formal instruction related to career development, with two-thirds receiving 3 or fewer hours of instruction. The fellows' major desire was for more information about transitioning their careers to faculty status (or practice). They rated clinical care and teaching as more important functions than did the faculty.
One surprising primary finding is that for junior faculty the amount of educational debt was not associated with future career plans when analyzed by a monthly debt/income method. This conclusion is supported by demonstrating that (1) the proportion of faculty with very high monthly payments is low, (2) faculty salaries are usually greater than $100,000/yr, (3) many faculty are in two-income families with the other person often in a high-salary position, (4) there was no association between amount of monthly educational debt payments and estimated likelihood of continuing in an academic career, and (5) research-oriented faculty and fellows had equivalent debt/income ratios as their counterparts without significant protected research time or interests. Although there undoubtedly are "outlier" scenarios of faculty with high educational debt payments and low institutional salaries who have non–wage-earning spouses and children living in an expensive metropolitan area, our data demonstrate that, for most faculty, educational debt payments are less than 5% of total household pretax income. In addition, women faculty are more likely to be in a two-income household and have lower educational debt pressures. These results are significant because they raise the possibility that current federal interventions to reduce educational debt will substantially affect the finances of only a small proportion of faculty who have salaries similar to physicians in this survey. We predict that, by itself, the NIH Loan Repayment Program will have only an overall modest effect on recruiting and retaining physician-scientists.
The faculty data may be biased in that the survey did not sample physicians that, based on their debt burden or pessimistic financial calculations during fellowship, never considered a faculty position and entered private practice before the date of the survey. However, the fellows' data (Figure 4) also shows no strong relationship between educational debt payment and academic career plans. In contrast, Figure 5 shows a relationship between lower debt/income ratio and greater likelihood of academic career for fellows (with the exception of the last category), but also shows that the range of the median monthly debt/income ratio across all the categories of academic likelihood is only 1.5%. We acknowledge that defining the finances of individuals is complex, and this survey was not designed to account for tax obligations, noneducational debt, spousal debt, or lifestyle expectations. On the other hand, because we asked respondents to calculate only their institutional income, it is likely that some fellows and faculty supplement their income with noninstitutional employment (moonlighting or consulting). For inpatient-based specialties, such as pulmonary and critical care medicine, noninstitutional income can be substantial. We did not specifically determine the total educational or individual debt of individual respondents or their households; future surveys could compare these different measures of debt and their association with career decision making. Career choices also may also be influenced by physicians' expectations of future income, which we did not estimate.
The fiscal situation of fellows is different from junior faculty. Fellows' overall debt/income ratio was almost three times higher than the faculty ratio and financial pressures likely account for the one-quarter of fellows who are in full or partial deferment of loan payments. However, this tight financial situation of the fellows is likely to be short-lived, because we show that taking a faculty position in a pulmonary or critical care division greatly reduces educational debt pressures and thereby may lessen the influence of debt on continuing in academics. It is uncertain how much fellows know of the financial and debt conditions of typical junior faculty members. It is uncertain whether the NIH Loan Repayment Program, which solely offers repayment of educational debt principal (i.e., does not immediately reduce monthly payments), has a large financial impact on attracting and retaining junior research faculty in pulmonary/critical care positions. There may be, however, other nonfinancial benefits to recipients of a Loan Repayment Award, such as professional recognition. Disseminating detailed financial data such as salary distributions, pay ladders, and average indebtedness after 5 or more years of faculty status may positively influence fellows' decisions to consider an academic career by alleviating concern over the degree of persistent indebtedness for junior faculty.
It is possible that some respondents in the "faculty" group were actually fellowship trainees who had been given the rank of instructor, as is the custom at some programs. This is unlikely to affect the overall faculty results substantially as instructors comprised only 14% of the faculty group. The 5-yr mean age difference between the faculty and fellow groups suggests that a large number of "older fellows" were not misclassified as "young faculty."
The mentor–protege relationship is multifaceted, and our results show that a relatively large number of persons (mean, 4.6) are perceived as "important" mentors from fellowship through assistant professorship. Mentoring for women faculty was similar to men: the major exception was the Mentor Advising scale, which assessed items related to promoting the protege's visibility and advice about academic progress. Both fellows and faculty rated highly their mentors' personal relationship with them, but mentors were more likely to advise trainees on their scientific work rather than on more global topics, such as academic or career progress. This discrepancy may be particularly important to junior faculty because the most highly desired information categories are academic issues of promotion, visibility, and balancing competing responsibilities. Mentors may not consider these areas of information relevant to the mentor–protege relationship, or these areas may be inherently more ambiguous than teaching the scientific method and reviewing data. Alternatively, mentors' own time and productivity pressures may place practical limits on the extent of discussion. Similarly, for fellows, the major area of hard-to-get information was how to advance to the next academic level. These results and comments from the focus groups suggest that developing physician-scientists very much desire and would benefit from an "academic roadmap" that addresses these topics.
Two analyses suggest that mentors' own academic training history influences the quality and preparation of their trainees. Both faculty and fellows who had mentors holding Ph.D. degrees had higher self-reported research skills and participated in more organized research or career development activities. These associations were weak, but consistent. The reason for this association cannot be determined from this survey design, but a logical explanation is that a mentor's graduate school experience (e.g., graded classes, examinations and defenses, thesis advisory committees, thesis writing) creates training standards that the mentor expects their M.D. trainees to meet. Undoubtedly, there are excellent M.D. mentors without Ph.D. degrees, but our findings suggest that incorporating specific graduate school training concepts into fellowship programs could increase the number of successful physician-scientists (11). This conclusion is somewhat tempered by recent reports that university-based research Ph.D. programs in both the sciences and the humanities have experienced development and retention problems similar to that of physician-scientists (12, 13). It is also possible that Ph.D. mentorship has no beneficial independent effects on the scientific skills of faculty or fellows, but that the association is confounded by highly motivated physicians' self-selection to train at institutions with strong graduate programs.
Annual reviews and meetings with supervisors have long served as formal mechanisms to assess academic progress, but peer normative data can also serve as an independent marker of achievement and career advancement. Although different institutions have varying expectations of research productivity as measured in publications or grant support, this survey suggests that less than one primary-author research publication per year represents average progress. About half of all publications by junior faculty are considered nonresearch publications. This fact should lead to further dialog among junior and senior faculty regarding the weight of nonresearch publications in academic career development and preparation for promotion decisions.
Job security was an issue repeatedly raised in the focus groups. It is logical that trainees who perceive their odds of sustaining a physician-scientist career as low would be less likely to commit their time and opportunity costs to train for that career. For instance, participants believed that their specialized research skills would be useless in a community practice job if they could not remain in academic medicine. Our finding that only 13% of faculty and 20% of fellows agreed that an academic career had more job security than a clinical practice position may explain much of the difficulties in attracting and retaining physicians in an academic career. Concerns about academic job security seem to have not abated since the survey of pulmonary and critical care junior faculty survey by Broaddus and Feigal in 1988 (14). An academic research career requires competing on a national level and even committed, well trained persons may be unsuccessful (15). However, there are few data available to trainees that allow them to make an accurate appraisal of the career success and retention rates for peer physician-scientists. Collecting and disseminating such data may offset some unduly pessimistic attitudes about the viability of an academic career in medicine. It would be very helpful for professional societies to develop methods to prospectively track the careers of a cohort of trainees and junior faculty. Similarly, tracking of the success rates and careers of junior grant awardees from the NIH, American Heart Association, and American Lung Association would be useful information for advisors, fellows, and junior faculty. Whether the solution to the issue of career uncertainty is to try to reduce it for faculty or to insulate trainees from this faculty concern is uncertain.
A limitation of this report is that the survey was focused on pulmonary and critical care–oriented physicians with respondents mostly practicing adult medicine. Although the survey touched on issues likely relevant to all trainees and junior faculty, it is unknown whether the specific results are generalizable to other subspecialties within pediatrics or internal medicine or to specialties such as surgery or psychiatry. Other variables that were not examined in this survey (the presence of children; spousal career nature; and personal or spousal expectations about financial stability) also may play roles in career decision making and would be important areas for future investigations. A response rate of 46% may create bias in the estimates of group characteristics, such as average age or monthly debt payments, but response bias should have a lesser effect on correlational analyses within the survey such as the relationship between debt payments and future career plans.
In summary, a trainee's decision to prepare and continue in a physician-scientist career has multiple determinants and incorporates personal characteristics, academic environment, work, and domestic lifestyle expectations, financial reward, and job security. If the future productivity of the nation's biomedical workforce and the mission of academic health centers is at least partially dependent on developing and retaining academic physician- scientist, then, based on our findings, we conclude that amelioration of educational debt is only one aspect of an effective solution. For fellows interested in an academic career, and for faculty, our data suggest that effective interventions to improve physician-scientist training and retention would include greater discussion of the academic career ladder and job security by mentors, formal career development meetings (supplemented by empiric data about productivity and timelines to grant success), processes to increase professional visibility, and implementing educational methods common to doctoral research programs. Identifying ways to reduce the anxiety and concern about career success in research for individuals who are committed to this pathway may also be an important method to alleviate this long-term problem. It is not known whether these interventions would be most effectively coordinated by divisional, departmental, school, or professional society organizations.
Acknowledgments
The authors appreciate the assistance of Todd Rockwood, Ph.D., in revising the survey format and content; Monika Simeonova and Elisha Malanga of the American Thoracic Society, Darlene Buczak of the Association of Pulmonary and Critical Care Program Directors and the Chest Foundation for assistance in compiling the survey list; Joan Watkins for transcription of the focus groups; and Courtney Broaddus, M.D., of University of California–San Francisco and Peter Bitterman, M.D., at the University of Minnesota for advice on survey content and analysis.
FOOTNOTES
Supported by a grant from the American Thoracic Society.
Versions of this research were published in abstract form in Am J Respir Crit Care Med 2002;165:A110.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.200503-325OC on September 22, 2005
Conflict of Interest Statement: C.R.W. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. J.B. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. R.R. has participated as a speaker in a conference that was partially financed by Ross, although the lectures were unrelated to their products. She owns stock in Abbot Laboratories as part of a portfolio. D.H.I. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
REFERENCES
National Heart, Lung and Blood Institute. Research training and career development: human resources for research in the 21st century: report of a November 1999 workshop. Bethesda, M.D.: National Heart, Lung, and Blood Institute. Available from: http://www.nhlbi.nih.gov/funding/training/train-rev.pdf (accessed November 12, 2004).
Lenfant C. Training the next generation of biomedical researchers: challenges and opportunities. Circulation 2000;102:368.
Rosenberg L. The physician-scientist: an essential—and fragile—link in the medical research chain. J Clin Invest 1999;103:1621–1626.
Ley T, Rosenberg L. Removing career obstacles for young physician-scientists: loan repayment programs. N Engl J Med 2002;346:368–372.
Zemlo T, Garrison H, Partridge N, Ley T. The physician-scientist: career issues and challenges at the year 2000. FASEB J 2000;14:221–230.
Palepu A, Friedman R, Barnett R, Carr P, Ash A, Szalacha L, Moskowitz M. Junior faculty members' mentoring relationships and their professional development in US medical schools. Acad Med 1998;73:318–323.
Kazzi A, Langdorf M, Brillman J, Handly N, Munden S. Emergency medicine residency applicant educational debt: relationship with attitude toward training and moonlighting. Acad Emerg Med 2000;7: 1399–1407.
Colquitt W, Zeh M, Killian C, Cultice J. Effect of debt on US medical school graduates' preferences for family medicine, general internal medicine and general pediatrics. Acad Med 1996;71:399–411.
National Institutes of Health awards to medical schools (medicine departments) fiscal year 2001. Available from: http://grants.nih.gov/grants/award/rank/medicine01.htm (accessed November 26, 2004).
The American Association for Public Opinion Research. Standard definitions: final dispositions of case codes and outcomes rates for surveys. Ann Arbor, MI: AAPOR; 2000.
Varki A, Rosenberg L. Emerging opportunities and career paths for the young physician-scientist. Nat Med 2002;8:437–439.
Golde C, Dore T. At cross purposes: what the experiences of doctoral students reveal about doctoral education. Report prepared for The Pew Charitable Trusts, Philadelphia, 2001. Available from: http://www.phd-survey.org (accessed November 12, 2004).
Council of Graduate Schools. Ph.D. completion and attrition: policy, numbers, leadership, and next steps. Washington, DC: Council of Graduate Schools; 2004.
Broaddus V, Feigal D. Starting an academic career: a survey of junior academic pulmonary physicians. Chest 1994;105:1858–1863.
Kotchen T, Lindquist T, Malik K, Ehrenfield E. NIH peer review of grant applications for clinical research. JAMA 2004;291:836–843.(Craig R. Weinert, Joanne Billings, Rita )