Characteristics of Pediatric Hospital Medicine Fellowship Program Directors

We developed and distributed a 15-question, cross-sectional national survey to members of the American Academy of Pediatrics-sponsored PHM PDs listserv. At the time of the survey, the listserv had 212 members. All active PDs and APDs are given access to the listserv; other listserv members are former or potential future PDs, or other educational leaders. Questions included, “What is your gender?” and “Is your race/ethnicity considered UIM (defined as: Black, Latinx, Pacific Islander/Native American)?” Additional questions included: Age, PD versus APD status, terminal educational degree, other educational training, duration in role and academic rank, completion of PHM or other fellowship, PHM board certification, full- or part-time status, and details about personal domain(s) of scholarship. Questions were pilot tested by a subset of the PHM PDs research subcommittee with expertise in survey design. Questions were then revised on the basis of feedback to improve clarity. The survey was distributed and data managed using Research Electronic Data Capture.⁹  The survey was open for 4 weeks in May 2021 and weekly reminders were distributed. Responses to questions were summarized using descriptive statistics. Only responses from PDs or APDs were included. The study was considered exempt by the institutional review board.

Responses were received from 56 current fellowship leaders (40 PDs and 16 APDs), with at least 1 leader responding from 43 of the 59 (73% of existing programs at the time of the survey) PHM fellowship programs. Our data included multiple respondents from the same program in 12 instances (11 programs had 1 PD and 1 APD respond, and 1 program had 1 PD and 2 APDs respond). Most respondents identified as female (71%) and were aged 50 or less years (80%; Table 1). Only 4 (7%) leaders self-identified as UIM.

Most respondents had received allopathic medical training, with only 2 respondents (3.5%) having osteopathic degrees. Half of surveyed leaders had received advanced nonmedical degrees: 27 had a master’s-level degree and 1 had both a master’s and Doctor of Philosophy degree. Degrees were most commonly in the fields of public health (n = 10, 33.3%), education (n = 8, 26.7%), or clinical research (n = 6, 20%) (Supplemental Table 2). Two respondents had a master’s in business health care administration. Many leaders had also pursued nondegree-granting additional training. The Academic Pediatric Association’s Educational Scholars Program and the Advancing Pediatric Excellence in Education program had been completed by 9 (16%) and 8 (14%) leaders, respectively, including 1 person who had completed both programs. Additional training cited included the Academic Pediatric Association’s Research Scholar’s Program (n = 1); the Association of Pediatric PDs Leadership in Educational Academic Development program¹⁰  or other leadership training (n = 2), clinical effectiveness, or quality improvement (QI) training (n = 2); and a variety of institutional or local education programs (n = 5).

About a third (n = 19, 34%) of respondents had completed a PHM fellowship, and an additional 21% (n = 12) had completed a non-PHM fellowship. The most frequent other fellowship reported was academic pediatrics (n = 4, 33%). Two leaders had completed an infectious disease fellowship, and 1 leader each reported graduating fellowship in critical care, neonatology, and environmental health. Two respondents had done research-based fellowships (Robert Wood Johnson and T32 Primary Care).

At the time of this survey, the ABP board exam for subspecialty certification in PHM had only been offered once. Forty (71%) leaders reported they had taken and passed this exam and were board certified. An additional 15 (27%) leaders were board eligible but not yet certified. One leader reported not being board eligible.

Most leaders were at the rank of assistant (n = 26, 46%) or associate (n = 22, 39%) professor, with only 6 (11%) at full professor, and had been at their current rank a mean of 5.47 years (averaged across ranks). The average time they had held the PD or APD role was 3.61 years. Only 4 leaders (7.1%) had a part-time work status.

Most leaders (n = 33, 58.9%) selected multiple domains of scholarship when queried about their primary domain for scholarly interest. When grouping by domain, the most represented was education (n = 37), followed by QI (n = 29), clinical research (n = 19), advocacy (n = 11), and other (environmental health, global heath, leadership n = 3). For program leaders who selected only 1 domain, the most common choice was education (n = 12), followed by QI (n = 5). Three leaders selected traditional hypothesis-driven clinical research as their sole domain of scholarship.

Finally, we stratified the results to compare PD responses to those of APDs (Table 1). Compared with PDs, APDs were still predominantly women (68.7% vs 72.5%), more frequently identified as UIM (12.5% vs 5%), and were younger (62.5% vs 42.5% <40 years old). A higher proportion of APDs had completed a PHM fellowship (56% vs 25%) and had advanced degrees (62% vs 45%). Frequency of selecting domains of scholarly interest was generally similar across the groups. Less APDs than PDs chose education as 1 of their domains (50% vs 72%), and APDs were slightly more likely to choose clinical research (47% vs 30%).

This national survey of PHM fellowship leaders confirms the high proportion of women serving as PDs or APDs and a low proportion identifying as UIM. In addition, almost half of PDs have not completed a fellowship themselves, although the proportion of APDs having completed a fellowship is much higher and may signal a coming shift in the field. Most survey respondents indicate that they have diverse scholarship interests, largely in education and QI, with a much smaller number in traditional hypothesis-driven clinical research.

Seventy-one percent of fellowship educational leaders responding to our survey were women, which approximates the gender makeup of the field as a whole.¹¹  This is a notably high proportion even when compared with other areas of pediatrics; pediatrics residency PDs are only about 50% women.⁴  However, recent studies showed that noneducational leadership roles in PHM, such as division or program leaders, remain at about 50% women. The reasons for this discrepancy are unknown, but factors such as lack of mentors,¹²  higher demands outside of work,¹³  and negative response when female physicians practice self-promotion¹⁴  have been reported. Further work to determine which of these or other factors specifically affect PHM and why is needed. Women in PHM are interested in educational leadership roles, as evidenced by this data, so increased representation in other domains of leadership may be achieved if faculty development, sponsorship, and QI projects are enhanced to identify and support future leaders. The creation of Advance PHM,¹⁵  a national collaboration to develop community, advocacy, and research around issues of gender equity, is 1 such example of how PHM is leading innovation in this space and can be an example for other pediatric and medical specialties. Transparency and focus on these issues can also help to promote PHM fellowship to women and allies interested in scholarship around these important topics.

Despite the successes relating to gender equity in PHM fellowship leaders, our data showed there remains limited UIM representation. Our finding that only 7% of PHM fellowship roles are filled by UIM faculty is not reflective of our patient population and merits ongoing measurement as efforts are undertaken to improve measures of diversity across our discipline. A multifaceted approach to increase diversity among trainees should include attention to the educational leadership charged with mentoring them.¹⁶  The high diversity of interests including education and QI of PHM fellowship leaders could potentially aid in the recruitment of UIM trainees. Broader and diverse interests likely improve the educational experience and offer flexibility. For example, a qualitative study of the experiences of UIM doctors in academic pediatrics identified the majority had areas of interest in education and advocacy.¹⁷  That same study identified mentorship from someone with similar interests and background, even if not UIM themselves, as a major driver for choosing the field of academic pediatrics. Our field must pursue actionable changes in recruitment, retention, promotion, and scholarship to further UIM representation.

Most PHM fellowships are only 2 years’ duration. This shortened time frame, coupled with the small number of PHM leaders in clinical research roles, is likely to serve as a barrier to completing traditional clinical research projects. QI projects or medical education research may be more likely completed with direct mentoring from fellowship leaders. Another challenge may be that most fellowship leaders are women and there have been longstanding gender disparities in grant funding, mentoring, and publication in traditional academia.¹⁸  In PHM, there remain gaps in representation at the senior author level, but there is increasing proportions of women publishing as first authors.¹⁹  Further research will be needed to see if this will eventually lead to improvements at all levels. Regardless, clinical research in PHM is likely to be necessary to move the field forward and a proportion of future hospitalists will need skills in conducting high-quality research. Pediatric emergency medicine serves as an instructive example where, over time, a group of skilled researchers were able to create a network (Pediatric Emergency Care Applied Research Network) to make significant contributions to the prevention and management of acute illnesses and injuries in children.²⁰  Similarly, the Pediatric Research in Inpatient Settings (PRIS) Network, which is a pediatric hospitalist research network,²¹  will need a steady influx of clinical researchers to advance the care of hospitalized children. PHM training programs will need to train future scientists, and the background of program directors is likely to be influential in the types of scholarship fellows choose to pursue. Previous work has demonstrated a strong correlation between the scholarly output of fellows and the scholarly output of program directors.^2,3  Additionally, some have called for shortening non-PHM fellowship training from 3 years to 2 years for those interested in pursuing clinical or clinician–educator careers suggesting that 3 years is felt to be necessary to develop clinician scientists.^2,3  Notably, several existing PHM fellowship programs have a mandatory third year to pursue additional scholarly training, and thus it is likely that a subset of PHM programs will yield a greater proportion of future researchers.¹⁵  Programs may choose to consider extending the fellowship an additional year for those interested in traditional clinical research, but this is likely to require additional faculty development. There may also be a role for novel approaches such as consideration of a network of national research experts to mentor fellows or further development of peer mentoring programs.²² 

Establishing a baseline of demographic information can be used to identify targets for or measure success of improvement measures. For example, in 2010, time since PD appointment for anesthesiology residency program directors averaged the same as in our study, about 4 years.²³  When the survey was repeated in 2018, those figures had not statistically changed, indicating high PD turnover, which led to initiatives attempting to promote job satisfaction.²⁴  PDs require a particular set of skills that includes excellence in education, administration, scholarship, and personnel management, as well as clinical competence. Previous literature has emphasized that PDs usually learn their job through trial and error without formal training.²⁵  High turnover and burnout among PDs have also been reported.⁴  Because PHM is still growing and changing as a field, our data provide a baseline that can be revisited over time. Future studies may reexamine this population in more detail around themes such as faculty development and wellness.

Limitations of our study include use of a survey which did not capture all current PDs, despite an excellent response rate. In addition, although we collected data on both PDs and APDs, the total number of APDs in the country is unknown and this limits the ability to draw conclusions from comparisons between those groups. As a cross-sectional demographic study, we can identify topics for further study but are limited in exploring the nuance and causality of factors we have identified. Future study may include interview-based exploration of themes related to gender, UIM status, and scholarship among our group of leaders.