The declining numbers of clinician-researchers in medicine, generally, and in psychiatry, in particular, has been the subject of increasing concern over the past two decades (1). There has been a growing literature proposing increased research training in medical school and residency as an important means of motivating young medical professionals (2, 3). A recent report by the Institute of Medicine (IOM), while acknowledging multiple factors as affecting recruitment into research, focused on residency training as a central issue (4). In addition to mandatory didactics in research methodology and enhanced research opportunities for residents, the IOM recommended reducing the clinical requirements for Board certification in psychiatry so as to allow interested residents to devote more of their time to research.

Although such proposals may seem plausible, there is little evidence to support their efficacy. The majority of recommendations for research training programs have been published without outcome data, and virtually none provide a means of separating the effects of motivation and innate ability from those of training. Thus, we do not know which parameters of training, or, indeed, what factors in general, motivate physicians to become researchers.

What residents can tell us about why they have or have not chosen research careers is a potentially useful, but unexplored, source of guidance in deciding where best to put our efforts in recruiting new researchers. In particular, we wished to investigate the influence of residency training versus other factors. To this end, we have conducted an exploratory survey of senior residents in large, research-rich, North American university psychiatry residency programs. We have identified a subgroup with high interest in research and compared them with residents with less such interest in terms of their research training, history of research experience, family issues, financial considerations, motivational factors, and plans after residency.

Our aim was not to conduct an epidemiological survey, nor to compare residency programs in recruitment of researchers, but to collect an enriched sample of residents with a high level of interest in conducting research as part of their careers, a group that comprises a small proportion of the total resident pool. We therefore contacted residency directors at U.S. psychiatry residency programs having 10 or more residents per class (total of 33 programs), as identified by data available from the National Resident Matching Program (NRMP), plus the University of Toronto (the largest, most research-intense program in Canada), asking them to forward a computer-based survey to senior residents in their general adult residency programs graduating in June 2008. We targeted large university programs because such programs are the most likely to have research-rich environments and attract residents with an interest in research. The survey, using Survey Monkey, consisted of 23 questions dealing with demographic information, educational background, financial issues, level of interest in research, research experience, research opportunities during residency, plans after residency, and factors enhancing or diminishing the attractiveness of research as an element of their careers. The questionnaire was devised by the authors, based on their knowledge and experience in psychiatric education. Responses were anonymous as to individual respondent as well as program. The project was certified as exempt by the IRB at Tufts Medical Center, Boston, MA.

We received confirmation of participation from 16 of the 33 residency programs, which sent surveys to a total of 189 graduating residents. Participating departments had mean NIMH funding for 2008 of $21,660,227 and a mean class size of 11 residents. Non-participating programs had mean NIMH funding of $16,548,631 and a mean class size of 11.8 residents. Participating programs were the following: Baylor, Emory, Johns Hopkins, Massachusetts General–McLean, University of California Irvine, UCLA, University of Illinois, University of Michigan, University of North Carolina, University of Pittsburgh, University of Texas Southwestern, University of Toronto, University of Virginia, University of Washington, Washington University, and Yale. We received 127 responses from residents, yielding a response rate of 67%. Returned questionnaires indicated consent to participate. We compared subgroups of respondents with differing levels of interest in research: those who said they had little or no interest in research (Low-Interest Group); those who indicated interest in collaborating on other people's projects, conducting industry-sponsored clinical trials, and/or being principal investigator on small-scale, low-budget studies, but did not want to be grant-funded principal investigators (Moderate-Interest Group); and those who may have endorsed any or all of the above types of research, but also said that they would like to be principal investigators on grant-funded research projects (High-Interest Group). We investigated differences in distribution of responses among the three groups by using the SPSS, Version 17.0.0, chi-square statistic function. Where appropriate, chi-square was used as a post-hoc test between pairs of groups to compare the Low-Interest Group separately with the Moderate-Interest and High-Interest Groups.

There were 37 residents (29%) in the High-Interest Group, 37 (29%) in the Moderate-Interest Group, and 53 (41%) in the Low-Interest Group. The groups differed markedly in post-residency plans: Whereas 54.1% of the High-Interest Group planned to enter research fellowships, only 5.4% of the Moderate-Interest Group and none of the Low-Interest Group did (χ² [2]=45.7; p<0.0001). Similarly, 67.6% of the High-Interest Group said they planned to spend more than 25% of their time engaged in research in their first post-residency position, whereas only 5.4% of the Moderate-Interest Group and none of the Low-Interest Group did (χ² [2]=67.3; p<0.0001).

In the total sample, 51.6% were women; 66.1% were married; 24.4% were parents; 21.4% had one or more researchers in their family; and 30.7% had master's or doctoral degrees in addition to medical degrees. The modal age range was 31–40 years, and the modal debt level was $100,000–$200,000. Table 1 compares demographic and background characteristics of those in the Low-, Moderate-, and High-Interest Groups. The gender distribution among the groups was markedly different, with the proportion of women decreasing from an overall level of 51.6% to a low of 24.4% in the High-Interest Group. Age, marital or parental status, and presence of researchers in the family of origin did not differ among the groups. However, level of debt did distinguish the groups. A minority of those in the High-Interest Group had total debt of more than $50,000, whereas progressively larger majorities of those in the Moderate-Interest Group and Low-Interest Group did. The likelihood of having a graduate degree in addition to a medical degree increased progressively with level of research interest.

The groups differed markedly in when they first considered research as part of a career choice (χ² [8]=45.9; p<0.001); 65% of those in the High-Interest Group first considered research before entering medical school (including 24% in high school). For the Moderate-Interest Group and Low-Interest Group, the comparable proportions were 44% and 27%, respectively. Earlier interest in research correlated significantly with greater expected time in research post-residency (Spearman rank-order correlation r=0.8; p<0.05).

As seen in Table 2, histories of research experience over the course of their education differed significantly among the groups. The groups did not differ in likelihood of having had research experience before medical school, with more than two-thirds of each group reporting research experience in college. Similarly, the modal time of first considering research for all respondents was in college. However, High-Interest Group members engaged significantly more in research than those in the other groups in both medical school and residency.

The residencies of all respondents presented similar research opportunities, but High-Interest Group respondents took advantage of them to a much greater degree than those in the other groups. An overwhelming majority of respondents in all groups reported that their residencies offered research tracks. However, very few of those outside the High-Interest Group had participated in such tracks. Similarly, likelihood of having a research mentor in residency, of devoting ≥5 hours weekly to research during some period of residency, and of designing one's own research project increased progressively with level of research interest. Of those in the High-Interest Group, 57.6% formally presented their project within their department; 42.4% presented at a national meeting; and 48.5% submitted a manuscript to a journal. The corresponding numbers from the Moderate-Interest Group and Low-Interest Group were progressively smaller, and too small to make statistical comparison meaningful.

In order to further investigate the impact of residency-training on recruitment into research, we looked separately at those respondents who first considered research during residency. Only 20 respondents (15.7%) fell into this category. Of those, 8 (40%) were in the Low-Interest Group; 7 (35%), in the Moderate-Interest Group; and 4 (25%), in the High-Interest Group. One resident (5%) was planning to enter a research fellowship, and 3 (15%) expected to spend more than 25% of time engaged in research in their first post-residency position.

Table 3 summarizes responses to questions about motivating factors for research for each group. Respondents were asked to indicate on a scale from –3 to +3 whether each of the factors shown on the table was positive, negative, or neutral as a motivator. The motivational roles of both intellectual and clinical interest were opposite for those with versus those without research interest. Both were seen as positives for those with high or moderate research interest, but were largely negative or neutral for the those with lower research interest. All groups viewed financial needs, family obligations, job availability, and job security as deterrents to research. However, job availability and security were slightly, but significantly, less likely to be viewed as deterrents for those with high interest in research. Influence from family of origin was largely neutral in all three groups, but was modestly, but significantly, more likely to be a positive motivating factor among those in the High-Interest Group. When asked about reasons for choosing their first post-residency position, those with high research interest were significantly less likely than other residents to rate financial needs (High-Interest Group: 27%, Moderate-Interest Group: 57%, Low-Interest Group: 66%; χ² [2]=13.8; p<0.001) or current family obligations (High-Interest Group: 27%, Moderate-Interest Group: 40%, Low-Interest Group: 47%; χ² [2]=33.1; p<0.001) as important. Among those in the High-Interest Group, the top five areas of research interest were neuroscience (61.8%), mood and anxiety disorders (41.2%), schizophrenia (32.4%), health systems (26.5%), and “other” (26.5%). The five least popular areas were women's mental health (5.9%), traumatic stress disorders (5.9%), cross-cultural psychiatry (5.9%), eating disorders (2.9%), and forensic psychiatry (0%).

This is the first study of which we are aware that asks psychiatry residents about factors that influenced their motivation to become researchers. At a level of 29% high research-interest residents, our sample is clearly “researcher-enriched,” as compared with psychiatric residents as a whole, an intentional result of our methodology. The validity of this group as comprising true prospective researchers is supported by their heavily research-oriented post-residency plans, track records of involvement in research from college (and, in the case of over 20%, from high school) through residency, and likelihood of having graduate degrees in addition to medical degrees.

Our data suggest that the great majority of respondents had already decided how interested they were in research by the beginning of residency, and few of those with less than the highest level of interest further explored research by taking advantage of the rich resources offered by their programs. Senior residents planning to be career researchers had career trajectories different from those with less research interest. Among those wishing to have research careers, over 86% first considered such a career before residency, usually in high school or college, and they participated in it in progressively higher numbers as their education proceeded. Most of those with lower research interest had also had research experience in college, but were not motivated by it to continue engaging in research. Fewer of them did so during medical school and fewer yet during residency. In sum, research opportunities in residency do not appear to be a major factor in recruitment of new researchers.

Because we did not explicitly ask respondents about their research interest at the beginning of residency, but only toward the end, we cannot be sure whether there were prospective researchers who had been discouraged by their residency experiences. However, of those in our Moderate-Interest Group who did have research mentors, over 82% said that they were highly helpful, providing no evidence that low quality of residency research experience discouraged them from setting more ambitious research goals.

Our results are consistent with the few previous studies that bear on the question of motivation for research among physicians. A study of pediatricians found that the most salient factors predisposing to research were a preexisting interest and having engaged in research during summers, undergraduate years, or postdoctoral research fellowships (5). As in the present study, clinical interest was listed as an incentive for research, and financial considerations were seen as a deterrent. A survey of child psychiatrists similarly found that individual motivation was a salient predictor of survival as a researcher (6). A study of medical student education found that students' level of research activity, but not whether such activity was required versus elective, predicted ultimate research productivity, suggesting high salience of intrinsic motivation (7). Our finding that higher research interest is related to lower debt is provocative, but difficult to interpret. Those with larger debt may be more likely to reject research careers because of financial needs. Alternatively, there may be common factors that make both lower debt and high research interest more likely. Further investigation in this area would be useful.

Our data indicating that clinical interest is a positive motivating factor for researchers is notable. Others have found that M.D.-Ph.D. researchers' careers are not adversely affected by time spent in clinical training and that researchers do not necessarily view residency as the best time to have gotten their formal research training (8). Conversely, longevity of grant funding among clinical researchers has been found to correlate with time spent doing research during medical school and research fellowships, but not during residency (9). Intense involvement with clinical work may impede research after residency (10) but reinforce interest in it during residency. Taken together, these findings suggest that curtailing clinical training requirements is not likely to enhance recruitment into research, and may even be counterproductive. This suggestion, like others targeting residency training as a major means of recruiting researchers, should be approached with caution.

The scarcity of women in our group of prospective researchers is one aspect of the more general difficulty that women face in establishing and maintaining academic psychiatric careers, as has been pointed out by others (11). The gender imbalance in the High-Interest Group may also account for the very low level of interest in conducting research in women's mental health and conditions such as eating disorders and traumatic stress disorders that are highly prevalent in women. Availability of high-quality mentorship may be especially important in fostering the careers of women researchers (12). Ensuring such mentorship, as well as understanding other factors that discourage women from engaging in research must be a prominent aspect of ongoing efforts to increase our pool of researchers.

Our conclusions must be tempered by the limitations or our study. First, the results are based on a small number of respondents, and it is impossible to know how many of them will actually go on to research careers. Second, we have collected detailed information about the research activity only of residents. It remains possible that more intensive or closely-mentored research experiences in college or medical school would be more effective in recruiting new researchers. Third, only about half of targeted programs participated in the study, leaving the possibility that respondents may not have been representative of the entire group. However, programs that did and did not participate were academic programs with quite similar size and amount of NIMH research funding, suggesting no apparent differences in type of residents they would attract. Finally, we must note that we have not corrected for the number of chi-square tests performed. Standard correction methods assume independence of the variables being tested, and it is very unlikely that answers to the questions on our survey are all independent. However, using even the most conservative application of a Bonferroni correction yields a minimum significance level of <0.002, which would leave significance of all our main conclusions intact.

Our study suggests that relatively early developmental factors may be the most important determinants of the pool of potential clinician-researchers. Efforts to restore clinical researchers to their earlier numbers should focus on elucidating key developmental influences, understanding and alleviating impediments specific to women, reducing the debt load of medical graduates, and allocating funding adequate to ensure sustained and secure careers for talented researchers.

The authors thank Professor Michael Lipton, Northeastern University, Boston, MA, for his help with data analysis, and the residents and residency directors who participated in the study.