Mentoring in Academic Medicine:  A Systematic Review FREE

Medical schools and residency and fellowship programs are charged with training health care professionals and with advancing clinical care, research, and education.^1- 2 Mentoring has been considered to be a core component of the duties of medical school faculty to facilitate successful fulfillment of this academic mission. It has been recognized as a catalyst for career success, and mentoring relationships have been cited as important in facilitating career selection, advancement, and productivity.^3- 4 However, mentor-mentee relationships are challenged by increased clinical, research, and administrative demands.^3- 4 Moreover, mentorship is often undervalued by academic institutions.⁵

To enhance the development of mentorship within academic institutions and to prevent further erosion of these vital relationships, it is important to understand the effect of mentorship on the mentees (and mentors), the variables associated with mentoring success, and the impact of mentoring interventions on career satisfaction and productivity. The purpose of this systematic review was to evaluate the evidence about the prevalence of mentorship and its effect on career development.

Relevant studies were identified by searching the following databases: (1) all EBM Reviews on Ovid-ACP Journal Club (1991-March/April 2006), Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effects, and Cochrane Central Register of Controlled Trials (1st Quarter 2006); (2) Ovid Current Contents, all editions (July 4, 1993–May 14, 2006); (3) Ovid PsycINFO (1967-May 7, 2006); (4) Ovid MEDLINE (1966-April 30, 2006); and (5) Scopus, an Elsevier abstract and citation database (1996-May 14, 2006). To increase the sensitivity of the search strategy, we searched MEDLINE using the term Mentor. Other databases were searched using the following key words or their combinations: Mentor; Mentoring; Medicine; Career Mobility; Leadership; Teaching; Preceptorship; Interpersonal Relations; Students; Research; Schools, Medical; Academic Medical Centers; Education, Medical; and Faculty, Medical. To identify additional studies, we searched the bibliographies of those studies found by electronic searches, contacted experts in the field for potential unpublished studies, and completed a manual search of relevant library journals. There were no language restrictions.

We identified all studies evaluating the impact of mentoring on career choices and academic advancement among medical students, residents, fellows, and staff physicians. We included all study designs except qualitative studies. For this study, mentorship was defined as “a dynamic, reciprocal relationship in a work environment between an advanced career incumbent (mentor) and a beginner (protégé), aimed at promoting the development of both”.⁶ The definition included distance mentorship. We did not include studies evaluating the impact of role models, who were defined as persons “who serve as a model in particular behavioral or social role for another person to emulate.”⁷

Two of the authors independently reviewed the titles and abstracts of retrieved publications and selected relevant articles for possible inclusion in the review. In the case of disagreement, the third author was consulted and a decision was made by consensus of all authors. In cases of doubt, full-text articles were retrieved for review and discussion.

Minimum inclusion criteria were a description of the study population and availability of extractable data. Two of the authors independently reviewed all full-text articles that met these criteria. The agreement of the raters was very good (κ = 0.78). A data collection form was used to extract study type, intervention, setting, participant demographics, and outcome measures. Disagreements in assessment and data extraction were resolved by consensus of all authors.

Since most included studies were surveys with heterogeneous measurements, statistical pooling of the results or assessment of publication bias was not possible. Instead, we tried to discern areas in which the impact of mentorship has been found, and to provide a narrative description of the results using a strategy suggested by the Best Evidence Medical Education Collaboration⁸ and based on the validity of the individual studies. Study quality was assessed on the basis of study design, validation of survey questionnaires, sample size and sampling frame, response rate, and outcome measures.

Two authors developed a categorization of themes arising from the study results, and independently assigned the studies to these defined categories. Where possible, the association between the mentorship and academic or professional choices was calculated as the odds ratio (OR) and its 95% confidence intervals (CI), using MedCalc version 8.0 (MedCalc Software, Mariakerke, Belgium).

We retrieved 3640 citations from the literature search. Review of abstracts led to retrieval of 142 full-text articles for assessment, and 42 articles were subsequently identified for inclusion in the study (Figure). Original data were available on 39 studies, described in 42 articles^9- 50; 2 studies were reported in 5 published articles^40- 44 (Table 1). Most of the studies (n=33) were performed in the United States. Among these, 2 included respondents from Canada,^15,32 and 1 from Puerto Rico.²⁶ Three studies were performed exclusively in Canada,^18,49- 50 2 in Great Britain,^17,49 and 1 in Germany.⁴⁸ The design of 34 (87%) of the 39 studies was cross-sectional survey, with response rates ranging from 5% to 99%. Three studies were before and after case series,^10- 12 1 was a case-control study nested in a survey,^42- 44 and 1 was a cohort study.⁹

Many of the studies had methodological limitations. Twelve studies reported details on survey development or testing. The cohort study had a small number of participants, unaccounted crossover between the groups, and large loss to follow-up, which may have affected the validity of the results. The nested case-control study was performed within a self-reporting survey, with a 65% response rate.

Only 5 studies provided details on how the mentorship relationship was formed.^{9,11- 12,22,45} Two studies described voluntary mentorship programs in which mentors were selected by mentees,^9,11 and 1 study described a program with a formal arranged mentorship relationship.¹² A survey of obstetrics/gynecology fellows showed that both the mentor and the mentee initiated most of the clinical mentoring relationship.²² Of 279 child and adolescent psychiatrists, 117 (42%) reported being assigned a mentor, 86 (31%) reported requesting a specific mentor, and 75 (27%) described independently initiating the mentor-mentee relationship.⁴⁵

Fifteen studies examined the prevalence of mentorship among medical students and physicians (Table 2).^{13,15- 16,19,22,29,31- 32,35- 36,38- 39,42,45,49} The prevalence ranged from 19% of adolescent medicine faculty who reported currently having a mentor¹⁹ to 93% of primary care research fellows who reported having a mentor.⁴² In 1 study that focused on the prevalence of mentorship at the undergraduate level, 36% of the third- and fourth-year medical students reported having a mentor.¹³

Four studies described the general importance of mentorship (Table 3).^{19,25,45- 46} Of surveyed child and adolescent psychiatrists, 16% identified mentors as the most important aspect of their training experience.⁴⁵ Among adolescent medicine faculty, 95% of the respondents described their mentor as important to them.¹⁹

Perceived importance of mentorship was related to career satisfaction. In a study of maternal/fetal medicine fellows in the United States,³⁸ the presence of a mentor was associated with satisfaction with their fellowship (OR, 5.83; 95% CI, 2.42-14.08). In a survey of faculty from 24 US medical schools,³¹ faculty members with mentors had significantly higher career-satisfaction scores than those without mentors (mean score, 62.6 vs 59.5 on a scale range from 20-100; P<.003).

Eight studies reported the influence of mentorship on personal development and career guidance.^{11- 13,16,22,45,49- 50} Five studies found that mentors were seen as an important career-enhancing factor for medical students, fellows, and staff physicians in various disciplines (Table 4).^{13,16,22,45,50} A study of Canadian obstetrics/gynecology fellows found that those who reported they had a mentor were more likely to achieve a promotion (hazard ratio, 2.33; 95% CI, 1.36-3.99).⁴⁹

Two studies described the effect of academic mentoring programs on professional development.^11- 12 Illes et al¹¹ assessed a mentoring program for radiology junior faculty at the Stanford University School of Medicine in their before and after case series. The program was voluntary with formal mentoring meetings every 6 months. Participants rated their satisfaction with mentoring meetings and the relative importance of major professional issues that emerged in their discussions (Table 4). Annual review of junior faculty performance in the areas of research, teaching, and patient care showed improvement in 52% of program participants in research, 26% in teaching, and 6% in patient care from first monitoring meeting to either promotion or end of study.¹¹ However, no control group was available for comparison. Wingard et al¹² evaluated a structured mentoring program for junior faculty at the University of California San Diego in a before and after study. The program was multifaceted and included professional development workshops, career planning, counseling sessions, formal mentoring, and community network building. The program significantly increased self-assessed confidence in participants' academic roles and skills in several areas including professional development, education, and administration, with increase in self-efficacy scores of 52%, 33%, and 76%, respectively.

Nine studies described the impact of mentorship on specialty choice, 4 on academic career choice,^27,32,35,38 and 2 studies focused on retention in academic medicine (Table 5).^{9,12- 14,21,24,26- 27,30,32,34- 35,37- 38,46} Mentorship was reported to be an influential factor in the selection of specialty. Respondents working in academic medicine rated the importance of the mentor in their career choices higher than respondents working in nonacademic primary care settings (mean score 2.36 vs. 1.82 on a 5-point scale; P<.001).³⁷

Four studies explored the relationship between mentorship and the mentees' interest in entering academic medicine. Pearlman et al³² found a significant correlation between the presence of a mentor and a plan to enter academics among neonatal/perinatal fellows (P=.01). In a study of the US maternal/fetal medicine fellows,³⁸ the presence of a mentor was associated with a fellow's desire to enter academic practice (41.8% vs. 21.5%; calculated OR, 2.81; 95% CI, 1.21-6.51). However, expectation about future practice type among US gynecologic oncology fellows was not associated with having a clinical or research mentor.³⁵ Miller et al²⁷ found that whether entering academics or private practice, dermatology micrographic surgery fellows placed equivalent importance on the influence of mentorship from the fellowship director on their career choice.

Two studies explored the association between mentorship and faculty retention. Benson et al⁹ reported on a 2-tiered program consisting of 1 year of preceptoring with the goal of orienting new faculty, and mentoring for junior faculty who had been with the organization for at least a year. The study showed that 38% of junior faculty who did not form preceptoring partnerships left the organization, compared with 15% of those who formed preceptoring partnerships (P = .12).The report did not provide any data on the retention of those who formed mentoring partnerships. At the University of California San Diego, 85% of mentoring program participants remained at their home institution, and 93% remained in academic medicine,¹² but there was no control group available for comparison.

Twenty-one studies described the impact of mentoring on research development and productivity. An apparent effect of mentoring was observed on research career guidance, productivity, and success (Table 6).^{11- 13,17- 18,20,23,31- 33,35- 36,38- 41,43- 47} Mentors increased mentees' self-confidence¹² and provided support and resources for research activities.^13,31,45 Respondents who had a mentor were more likely to allocate more time to research^23,31,44; they were more productive in research in terms of number of publications and grants,^{11,23,32,35,43- 44} and were more likely to complete their thesis.³⁸ Lack of mentorship was identified as a specific barrier to completing scholarly projects and publication.^17,36,39 A survey with a nested case-control study found an association between having a mentor and having a research grant as a principal investigator (OR range, 2.1-3.1).^43- 44 The influence of a mentor was an important motivating factor in pursuing research training or career.^{18,20,33,40,47} Research fellows who had had a mentor were more likely to provide mentorship to others (multivariate OR, 8.9; 95% CI, 1.8-42.4).⁴⁴

Three studies explored mentorship experiences of women physicians,^15,23,48 6 studies explored differences between sexes in the mentorship experience,^{13,16,29- 31,49} and 1 study evaluated an intervention to eliminate some of these differences.¹⁰ A survey of third- and fourth-year medical students at the University of California San Francisco¹³ found that 40% of men and 33% of women had mentors (calculated OR, 1.32; 95% CI, 0.77-2.27). Graduating students from the same school rated having a research mentor as the most important factor that influenced their specialty choice (1.95 on a 5-point scale ranging from 1 [very important] to 5 [unimportant]), but there was no difference between men and women.³⁰

In a survey of medical students, housestaff, fellows, and junior faculty at the University of California San Francisco,²⁹ 22% of women junior faculty and 21% of women on housestaff had never had a professional mentor; the same was true for 9% of men junior faculty and 16.5% of men on housestaff. There was no mentor reported in their current position at the university for 43% of the housestaff (same for men and women) and 45% of the women junior faculty; the result for men junior faculty was not given. Men were 3 times as likely as women to describe a relationship with a mentor as a positive experience that influenced their careers. Negative experiences most often mentioned by both sexes were lack of funding and lack of a mentor: 24% of the women identified the lack of a mentor as 1 of the 2 most negative experiences they had in their careers.

Coleman et al¹⁶ explored differences in perceptions of mentoring by surveying US obstetrics/gynecology residents by race and sex in a survey study. White women reported that they did not currently have a mentor more often than any other group of residents (59.8% vs 68.1% reported by white men; P< .001). Among Hispanic and African American residents, men reported more active and consistent advising than women (30.3% vs 27%; P value not stated).

There were some differences by sex among faculty in perception of the impact of mentorship on success. In a survey of obstetricians/gynecologists on Canadian medical faculties,⁴⁹ women were more likely than men to indicate that they perceived a lack of a mentor to be a barrier to their promotion (42% vs 24%; P< .001), although there was no difference by sex in prevalence of having a mentor (42% of women vs 46% of men). Having a mentor was associated with a higher likelihood of promotion to professor (HR, 2.33; 95% CI, 1.36-3.99). However, a study of US women faculty aged 50 years or younger did not find a correlation between having a mentor during training and academic rank.²³ Women pediatric surgeons in both the United States and Canada identified lack of appropriate mentorship as a major obstacle to a successful academic career (mean score, 2.71 [SD,1.17] on a scale of 1 [not important] to 4 [very important]).¹⁵ A similar finding was reported in a survey of US medical faculty³¹: more women than men believed that inadequate mentoring had impeded their career growth (48% vs 36%; P = .01). Lack of mentoring was also recognized in a survey of female academic surgeons in Germany, where 70% of respondents identified absence of mentoring programs as an obstacle in academic surgery, and 80% thought that better mentoring would improve the position of female academic surgeons.⁴⁸

The survey of US medical faculty³¹ found that mentors were predominantly white men, although women were more likely to have women mentors (23% vs 10%; P = .001). A similar result was reported by Coleman et al¹⁶ with the majority of mentors for both men and women residents being men, although women were significantly more likely than men to have a woman mentor (P< .001). These 2 studies had different findings about the importance of concordance of sex. In the study of faculty, 80% of the women reported that it was not important to have a mentor of the same sex, while in the study of residents, women were more likely than men to state that a same-sex mentor would be more understanding (41.4% vs 33.4%;  P<.001). Another study found that the mentor's sex was not a significant influence on either the number of publications or the percentage of time spent on research.²³

In a before and after case series, Fried et al¹⁰ described a multifaceted intervention to correct career obstacles based on sex that were reported by women faculty in the Department of Medicine at the Johns Hopkins University School of Medicine. Prior to implementation of the intervention, a faculty survey found that 44% of women and 59% of men expected to be promoted; 58% of women and 71% of men wanted to be in academic medicine in 10 years; 23% of women and 47% of men expected to be in academic medicine in 10 years; and 63% of women and 43% of men seriously considered leaving academic medicine (all P< .001). There were no differences by sex in prevalence of having a mentor. However, more women than men (32% vs 10%; P = .004) reported that their mentor used their work to advance their own career rather than that of the mentee. A 3-year intervention period followed the survey and included problem identification; leadership; education of faculty; and interventions to improve faculty development, mentoring, and rewards, as well as to reduce isolation and structural career impediments. Interventions were evaluated using a modified baseline questionnaire and found an increase in the percentage of women who had a mentor (31% vs 65%; P = .005) and expected to be promoted (44% vs 73%; P< .001), and a smaller percentage of women who seriously considered leaving academic medicine (63% vs 28%; P< .001). Both sexes reported that mentoring had improved (25% women, 22.5% men, difference not significant, exact P value not stated). Among men, the proportion who expected to be promoted increased from 59% to 76%. An increase in the proportion of men who expected to remain in academic medicine was also found, but it was smaller than in women (183% in women vs 57% in men).

To our knowledge, this is the first systematic review of the evidence of the relationship between mentorship and career choice, career progression, and scholarly productivity. The review of 39 studies reported in 42 articles revealed an absence of experimental research about mentoring, but it does outline current knowledge about mentorship. The available evidence showed that fewer than 50% of medical students and in some fields fewer than 20% of faculty members had a mentor. There was a perception that women had more difficulty finding mentors than their colleagues who were men. Mentorship was reported to be an important influence on personal development, career guidance, career choice, and productivity. Respondents identified mentoring to have an important effect on research productivity, including publication and grant success.

However, the poor quality of these studies does not allow conclusions to be made on the effect size of mentoring on any aspect of academic and professional development. Of the 39 studies, 34 (87%) were based on cross-sectional self-report surveys and did not utilize a comparison group without mentoring or with standard care. The median sample size of surveys selected for the review was 219 (range, 18-5624) and the median response rate was 62% (range, 5%-99%), with larger studies having smaller response rates. Many studies provided little detail on how the surveys were constructed or on the study sampling frame. The role of the mentor and content of mentorship greatly differed among the studies, ranging from an informal personal support to formalized mentorship relations. The majority of the studies did not mention if a mentor was assigned or self-identified. Moreover, none commented on how frequently mentors and mentees met or on the intensity of their interaction. There was little mention of potential adverse outcomes associated with mentoring other than one study that identified the perception that mentors used the mentees' work to advance their own career. All of the studies were completed in North America, the United Kingdom, and Germany, and may not accurately reflect developing and other countries.⁵ The limitations of this evidence preclude its use to suggest mentorship strategies that should be implemented at academic institutions.

Systematic reviews on the effects of mentorship in other fields, such as nursing⁵¹ and business,⁵² also show lack of valid evidence for the effectiveness of mentoring, indicating a general need for clarification of theoretical and conceptual perspectives in order to increase our knowledge of mentorship, particularly its traditional career and psychosocial functions. Understanding mentorship in medicine would benefit from assessing theories and evidence from other fields, such as social sciences, education, and business research.^53- 55

Two of the 4 intervention studies reported multifaceted interventions^9- 10 but it was unclear which elements had an effect on career advancement. Also, the studies in this review were not able to differentiate if the observed outcomes were the result of receipt of mentoring or the individual characteristics of the mentee. Management research has shown that personality characteristics can influence a person's likelihood of receiving mentoring.⁵⁶ Individuals with good internal control, high self-monitoring skills, and emotional stability were more active in seeking a mentoring relationship, which in turn contributed to receiving actual mentoring and career success.⁵⁶ Similar research is needed in medical settings to address the importance of personality traits in receiving and providing mentoring.

Despite the limitations of the evidence, some suggestions can be made regarding mentorship. Given that mentorship can have an effect on personal development, career choice, and research productivity, administrators, program directors, and mentors should encourage mentorship activities focusing on these areas. For example, guidance around research and access to relevant resources enhance productivity and should be regarded as key features of a mentorship relationship. It is not clear if mentors should be assigned or self-identified; this represents an area for future research. Mentorship should be available throughout training and career establishment, although different mentorship qualities may be required at these stages. Mentees should strive to find a mentor who can provide them with the required support for their career and personal development, including research resources where relevant. Efforts need to be made to ensure that mentorship opportunities are provided to women and individuals representing diverse ethnicities. However, it is not clear that mentors and mentees need to be of the same sex.

The results of this review provide an outline of common themes for future research: (1) the effect of mentorship on those interested in education-based careers; (2) the effect of strategies to enhance mentorship for women; and (3) the effects on career development and productivity of formal mentoring vs informal mentoring, personality and behavioral constructs, and multifaceted programs vs single component strategies. However, the quality of evidence does not allow practical recommendations to guide mentors in doing a better job and mentees in selecting a mentor. Research on the effects of mentoring on career choice and advancement must address contextual issues and use cross-disciplinary approaches and robust study designs, ideally including randomized trials. If it is not practical to randomize participants to a mentorship program vs usual practice, alternatives include randomizing to a multifaceted intervention or a single-component intervention. A prospective cohort study design could be used to identify those trainees or faculty with and without mentors and follow their cases forward to assess career choice and development, personality and social issues related to the mentorship process, and time requirements and costs of mentorship.

All of these study designs could be performed at single sites but would be more powerful if they were conducted across multiple sites. This would require collaboration under the leadership of the deans of medicine and organizations such as the Association of Professors of Medicine and other individuals and organizations interested in preserving academic medicine. Given the responsibility of medical schools and graduate programs for training health care professionals and for advancing clinical care, research, and education, these organizations should feel compelled to stimulate interest in mentorship and to evaluate such efforts. Education and faculty development initiatives should be subjected to the same valid forms of evaluation expected for interventions such as drug therapy.