Video Abstract

Video Abstract

Close modal
BACKGROUND AND OBJECTIVES:

The educational requirements for pediatric fellows include at least 12 months of scholarly activity and generation of a work product. Yet there lacks detailed guidance on how programs can best integrate scholarly activity training into fellowships. Our objectives were to understand the resources and barriers to training and identify factors associated with productivity.

METHODS:

We surveyed pediatric fellowship program directors (FPDs) nationally in 2019. Data analysis included descriptive statistics, χ2 and Fisher’s exact tests, and multivariable modeling to identify factors associated with high productivity (>75% of fellows in the past 5 years had an article from their fellowship accepted).

RESULTS:

A total of 499 of 770 FPDs responded (65%). A total of 174 programs (35%) were highly productive. The most frequent major barriers were a lack of funding for fellows to conduct scholarship (21%, n = 105) and lack of sufficient divisional faculty mentorship (16%, n = 79). The median number of months for scholarship with reduced clinical obligations scholarship was 17. A total of 40% (n = 202) of FPDs believed training should be shortened to 2 years for clinically oriented fellows. Programs with a T32 and a FPD with >5 publications in the past 3 years were twice as likely to be productive. Not endorsing lack of adequate Scholarship Oversight Committee expertise and a research curriculum as barriers was associated with increased productivity (odds ratio = 1.83–1.65).

CONCLUSIONS:

Despite significant protected fellow research time, most fellows do not publish. Ensuring a program culture of research may provide the support needed to take projects to publication. The fellowship community may consider reevaluating the fellowship duration, particularly for those pursing nonresearch focused careers.

What’s Known on This Subject:

Engagement in scholarly activities is an entrustable professional activity of pediatric subspecialists. Its importance is reflected in the training requirements for pediatric fellows. However, there lacks detailed guidance for fellowship programs on how to train fellows effectively in scholarly activity.

What This Study Adds:

In this national survey of pediatric fellowship program directors, we assess resources for and barriers to fellow scholarly activity training and factors associated with productivity. Productivity was associated with having a productive director, training grants, mentorship, and adequate core curriculum.

Engagement in scholarly activities has been recognized as an essential entrustable professional activity of a subspecialist.1,2  Although most pediatric fellows do not pursue physician-scientist careers, an understanding of research methodologies promotes competency in the critical evaluation of scientific literature and practice of evidence-based medicine.3  The importance of the scholar role is reflected in the fellowship training requirements set forth by the American Board of Pediatrics (ABP) and the Accreditation Council for Graduate Medical Education (ACGME). A minimum of 12 months of scholarly activity and of clinical training is required, allowing for flexibility in the remaining 12 months. All fellows must participate in a core curriculum in scholarly activities and complete a work product.4,5  Despite these requirements, detailed guidance on how to best develop a scholarly activity program is lacking.

The fellowship program director (FPD) plays an important role in ensuring a meaningful fellow scholarly activity training experience. FPDs must identify a mentor and Scholarship Oversight Committee (SOC) for each fellow and verify successful training completion.4,5  FPDs, however, often face challenges that impact fellow scholarly activity training, including high clinical demands for fellows, funding insecurity, and limited mentors because of the shrinking pipeline of pediatric subspecialty physician scientists.68  In a 2014 national survey of FPDs, researchers assessed their perspectives on fellow scholarly activity requirements but did not assess variability in program resources and perceived training barriers.9  There is no study to our knowledge in which researchers have tried to identify characteristics of training programs with high rates of fellow productivity, measured by traditional academic productivity metrics. Although there are multiple ways to define “success,” the Next Accreditation System has increased its focus on traditional outcomes such as presentations, publications, and grants.10  To address these gaps, with this study, we aimed to determine the resources for and perceived barriers to scholarly activity training in pediatric fellowship programs nationally and identify factors associated with high fellow productivity. Knowledge gained through this study can be used by FPDs to enhance their own program and by the ACGME or ABP to inform scholarly activity training structure and requirement guidelines.

We anonymously surveyed all current pediatric FPDs of ACGME-approved, 3-year fellowship programs from 14 subspecialties between March and October 2019. The subspecialty pediatric investigator network (SPIN) maintains this FPD database. SPIN is a medical education research network designed to facilitate the education and assessment of pediatric fellows.11 

We developed a novel electronic survey instrument (Supplemental Information) on the basis of our previous published work and comprehensive literature review.9,12,13  We obtained content validation and expert feedback from the SPIN steering committee, which consists of 1 to 2 FPD representatives from each subspecialty and experts in medical education, evaluation, and assessment. Cognitive interviews were conducted with 3 FPDs and 2 fellowship research directors at the lead author’s home institution. The revised survey was then pilot tested with 2 FPDs at the home institutions of each investigator and further revised. Pilot answers were not included in the final analysis, but we sent participants the final survey to allow for their participation.

The final survey consisted of 19 questions in 7 domains: (1) program characteristics, (2) scholarly activity engagement and work products, (3) resources, (4) barriers, (5) beliefs about fellowship training and scholarly activity, (6) satisfaction, and (7) FPD scholarships. Sixteen questions were closed-ended and three were open-ended.

We used the LimeSurvey platform for survey administration. An e-mail with the survey link was sent to all FPDs via the SPIN listserv, along with multiple reminders targeting nonresponders. FPDs were informed that participation was voluntary, the survey would take ∼10 minutes, no personal health information would be collected, and no incentives would be offered. Respondents could skip any question and review and change answers during survey completion. Completeness checks occurred after submission. Respondents were prevented from accessing the survey twice. We received Institutional Review Board approval from Weill Cornell Medicine.

We included all FPDs who reported having at least one fellow in the last 3 years. We defined hospital-based fellowships as neonatal-perinatal medicine, critical care, and emergency medicine. FPD survey responses were described as n (%), mean (SD), or median (minimum and maximum). We used the term “scholarly activity curriculum” to refer to the core curriculum in scholarly activities required by the ABP.14  We performed bivariate tests to identify program characteristics associated with perceived barriers. Our primary outcome for productivity modeling was whether >75% of fellows in the past 5 years had an article accepted in a peer-reviewed journal. Characteristics determined a priori to influence productivity included specialty type, program type (hospital based versus nonhospital based), number of fellows, percent of fellows pursuing clinical careers after graduation, adequate time provided for scholarly activity, adequate resources, barriers, and FPD beliefs. Characteristics were compared between productivity groups by using χ2 and Fisher’s exact tests or independent 2-sample t tests and Wilcoxon rank tests. A multivariable logistic regression modeling productivity was constructed with all characteristics significant in bivariate tests. A final model was derived through bidirectional stepwise selection on the basis of Akaike information criterion. Additional multivariable models were used to explore productivity, defined as >75% of fellows in the past 5 years presenting at a national or international meeting and >50% of fellows in the past 5 years having an extramural grant application submitted or accepted during fellowship. These models were built with all predictors used in the primary outcome model. We used similar methodology to explore factors associated with the belief that fellowship training should be shortened to 2 years for fellows interested in clinician or clinician-educator careers. All analyses were 2-sided, with significance evaluated at the .05 α level. Analyses were performed in R version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria).

We received responses from 499 of 770 FPDs, for an overall response rate of 65% (Table 1). We achieved at least a 50% response rate in all subspecialties except gastroenterology (response rate of 36%). The majority of fellowships had 1 to 3 (n = 187; 37.5%) or 4 to 6 fellows (n = 162; 32.5%), were nonhospital based (n = 338; 67.7%) and were affiliated with an academic medical center (n = 448; 89.8%; Table 2). Geographic representation was diverse. There was no significant difference between responding and nonresponding programs with regard to size or geographic region, although a lower proportion of community programs participated.

TABLE 1

Response Rate by Subspecialty

Subspecialtyn of Total (%)
Adolescent medicine 20 of 27 (74.07) 
Cardiology 36 of 59 (61.02) 
Child abuse 15 of 27 (55.56) 
Critical care 49 of 65 (75.38) 
Developmental-behavioral 28 of 38 (73.68) 
Emergency medicine 58 of 78 (74.36) 
Endocrinology 39 of 68 (57.35) 
Gastroenterology 22 of 61 (36.07) 
Hematology-oncology 46 of 72 (63.89) 
Infectious diseases 41 of 59 (69.49) 
Neonatal-perinatal medicine 54 of 100 (54) 
Nephrology 31 of 36 (86.11) 
Pulmonology 33 of 48 (68.75) 
Rheumatology 27 of 32 (84.38) 
Overall 499 of 770 (64.81) 
Subspecialtyn of Total (%)
Adolescent medicine 20 of 27 (74.07) 
Cardiology 36 of 59 (61.02) 
Child abuse 15 of 27 (55.56) 
Critical care 49 of 65 (75.38) 
Developmental-behavioral 28 of 38 (73.68) 
Emergency medicine 58 of 78 (74.36) 
Endocrinology 39 of 68 (57.35) 
Gastroenterology 22 of 61 (36.07) 
Hematology-oncology 46 of 72 (63.89) 
Infectious diseases 41 of 59 (69.49) 
Neonatal-perinatal medicine 54 of 100 (54) 
Nephrology 31 of 36 (86.11) 
Pulmonology 33 of 48 (68.75) 
Rheumatology 27 of 32 (84.38) 
Overall 499 of 770 (64.81) 
TABLE 2

Characteristics Between Nonresponding and Responding Programs

Nonresponding (n = 271)Responding (n = 499)POverall (n = 770)
Region, n (%)     
 Midwest 71 (26.2) 124 (24.8) .415 195 (25.3) 
 Northeast 82 (30.3) 130 (26.1)  212 (27.5) 
 South 80 (29.5) 157 (31.5)  237 (30.8) 
 West 38 (14.0) 88 (17.6)  126 (16.4) 
No. of fellows,a median (minimum–maximum) 4 (1–20) 5 (1–24)  5 (1–24) 
No. of fellows, n (%)     
 1–3 122 (45.0) 187 (37.5) .085 309 (40.1) 
 4–6 89 (32.8) 162 (32.5)  251 (32.6) 
 7–9 35 (12.9) 83 (16.6)  118 (15.3) 
 >9 25 (9.2) 67 (13.4)  92 (11.9) 
Setting, n (%)     
 Military 0 (0) 9 (1.8) <.001 9 (1.2) 
 Community 25 (9.2) 9 (1.8)  34 (4.4) 
 Academic 244 (90.0) 442 (88.6)  686 (89.1) 
 Other 2 (0.7) 39 (7.8)  41 (5.3) 
Nonresponding (n = 271)Responding (n = 499)POverall (n = 770)
Region, n (%)     
 Midwest 71 (26.2) 124 (24.8) .415 195 (25.3) 
 Northeast 82 (30.3) 130 (26.1)  212 (27.5) 
 South 80 (29.5) 157 (31.5)  237 (30.8) 
 West 38 (14.0) 88 (17.6)  126 (16.4) 
No. of fellows,a median (minimum–maximum) 4 (1–20) 5 (1–24)  5 (1–24) 
No. of fellows, n (%)     
 1–3 122 (45.0) 187 (37.5) .085 309 (40.1) 
 4–6 89 (32.8) 162 (32.5)  251 (32.6) 
 7–9 35 (12.9) 83 (16.6)  118 (15.3) 
 >9 25 (9.2) 67 (13.4)  92 (11.9) 
Setting, n (%)     
 Military 0 (0) 9 (1.8) <.001 9 (1.2) 
 Community 25 (9.2) 9 (1.8)  34 (4.4) 
 Academic 244 (90.0) 442 (88.6)  686 (89.1) 
 Other 2 (0.7) 39 (7.8)  41 (5.3) 
a

Two programs were excluded for the question on the number of fellows because of presumed order entry errors, with reports of 111 and 813 fellows in their individual specialty program.

Twenty-one percent (n = 105) of FPDs reported that >25% of their graduating fellows over the past 5 years pursued nonacademic careers (100% clinical careers in private practice or industry). Another 194 (24%) reported >25% of their graduating fellows in the past 5 years pursued a 100% academic clinical career. Nearly 60% (n = 286) reported that at least some graduates pursued academic careers with a focus in medical education, and 32% (n = 159) reported at least some graduates pursued academic careers with a focus in quality improvement (QI) and/or safety.

There was limited variability in the type of scholarly activity completed by graduating fellows over the last 5 years. Clinical research was most common, with 54.1% of FPDs (n = 270) reporting it was the type of scholarly activity completed by at least one-half of graduating fellows, including 174 (34.9%) who reported clinical research was completed by >75% of graduates.

Other types of scholarly activity allowed by the ABP were much less frequently proposed. One-half of FPDs (n = 250) reported “none” for the percentage of graduating fellows over the past 5 years for whom translational research was undertaken, and 211 (42.3%) reported none for basic science research. The results were similar for QI (42.7%; n = 213) and pursuit of an additional degree (n = 242; 48.5%). The number of FPDs reporting “none” was even higher for other types of scholarly activity: 320 (64.1%) for clinical care guideline development, 327 (65.5%) for medical education research, 364 (72.9%) for health services and policy research, and 427 (85.6%) for bioethics research.

Resources to support fellow scholarly activity training were variable across programs (Fig 1). The most common resources were funding for fellows to present their work at national meetings (n = 469; 94.0%) and statistical support (n = 442; 88.6%). The least common resources were a T32 training grant (n = 149; 29.9%) and funding to support an extra research year (n = 114; 22.8%).

FIGURE 1

Resources available across programs to support fellow scholarly activity training. SA, scholarly activity.

FIGURE 1

Resources available across programs to support fellow scholarly activity training. SA, scholarly activity.

Close modal

Most programs had only a common departmental research curriculum (n = 244; 48.9%). Approximately 17% (n = 84) had a division-specific research curriculum, 24.4% (n = 122) had both, and 9.8% (n = 49) had no curriculum.

There was wide variability in the number of months allocated for scholarly activity that were free of or had reduced clinical obligations (such as having only continuity clinic or call), both across programs and subspecialties (Fig 2). The median number of months was 17 (range: 0–29). One-quarter of programs (n = 137) had at least 21 months free of or with reduced clinical obligations. There was a statistically significant difference between hospital-based and nonhospital-based fellowships (medians of 15 and 18 months respectively; P < .001).

FIGURE 2

Time allocated for scholarly activity training with free or reduced clinical obligations by subspecialty discipline.

FIGURE 2

Time allocated for scholarly activity training with free or reduced clinical obligations by subspecialty discipline.

Close modal

The majority of FPDs believed their resources to support fellow scholarly activity training were either extremely or very adequate (n = 115 and 23.0%; n = 244 and 48.9%). One-quarter (n = 131) of FPDs described their resources as only somewhat adequate and 1.8% (n = 9) characterized them as not at all adequate. The vast majority of FPDs felt the time allocated for scholarly activity was sufficient (n = 477; 95.6%).

FPDs identified many major and minor barriers (Fig 3). The most significant major barrier was lack of funding for fellows to conduct scholarly projects (n = 105; 21.0%), followed by lack of sufficient divisional faculty mentorship (n = 79; 15.8%) and competing clinical responsibilities (n = 75; 15.0%; Fig 3). The barriers least reported were lack of adequate SOC expertise and lack of sufficient institutional faculty mentorship, characterized as “not a barrier” by 67.3% (n = 336) and 71.3% (n = 356) of FPDs, respectively. In bivariate analyses, lack of sufficient divisional, departmental, and institutional faculty mentorship were more likely to be perceived as minor or major barriers by programs in which >25% of fellows pursued nonacademic careers and by hospital-based programs (P < .01 for each barrier). Lack of adequate expertise on the SOC was also associated with hospital-based programs (P < .001), and lack of sufficient faculty mentorship at the divisional level was more of a major or minor barrier for western region programs (P = .04).

FIGURE 3

Barriers across programs to complete scholarly activity.

FIGURE 3

Barriers across programs to complete scholarly activity.

Close modal

The vast majority of FPDs reported that all graduating fellows within the past 5 years completed a scholarly project (n = 477; 95.6%), although success in achieving traditional productivity metrics was mixed. The most common accomplishment was >75% of fellows in the past 5 years presenting at a national or international meeting (n = 309; 61.9%). Only 34.9% (n = 174) reported that >75% of their fellows over the past 5 years had an article accepted in a peer-reviewed journal based on work completed during fellowship. Percentages were highest in infectious diseases (n = 22; 53.7%), cardiology (n = 18; 50%), and endocrinology (n = 19; 48.7%). Grant submission or acceptance was the least commonly achieved metric, reported by only 66 FPDs (13%) for >50% of their fellows in the last 5 years; 172 (34.5%) reported that no fellows had achieved that metric.

We identified 4 factors from multivariable modeling associated with productivity (Table 3), defined as programs in which >75% of fellows over the past 5 years had an article accepted based on work completed during fellowship (174 of 499 programs): (1) having a T32 training grant (odds ratio [OR] = 2.13; confidence interval [CI]: 1.41–3.24); (2) the FPD having >5 publications in the last 3 years (OR = 1.88; CI: 1.27–2.80); (3) not endorsing a lack of adequate expertise on the SOC as a barrier (OR = 1.83; CI: 1.14–2.98), and (4) not endorsing a lack of an adequate core research curriculum as a barrier (OR = 1.65; CI: 1.07–2.56). For multivariable modeling associated with >75% of fellows in the past 5 years presenting at a conference, the only factor that was significant was the FPD having >5 publications in the last 3 years (OR = 1.58; CI: 1.08–2.30). For >50% of fellows who had a grant submitted or accepted, the only factor that was significant was having a T32 training grant (OR = 5.99; CI: 3.39–10.88).

TABLE 3

Multivariable Logistic Regression Modeling High Productivity for Fellows Publishing Articles

Adjusted OR (95% CI)
T32 training grant (versus no T32) 2.13 (1.41–3.24) 
FPD with >5 publications in the last 3 y (versus ≤5 publications in the last 3 y) 1.88 (1.27–2.80) 
Lack of adequate expertise on the fellowship SOC is a not a barrier (versus a minor or major barrier) 1.83 (1.14–2.98) 
Lack of an adequate core research curriculum is not a barrier (versus a minor or major barrier) 1.65 (1.07–2.56) 
Adjusted OR (95% CI)
T32 training grant (versus no T32) 2.13 (1.41–3.24) 
FPD with >5 publications in the last 3 y (versus ≤5 publications in the last 3 y) 1.88 (1.27–2.80) 
Lack of adequate expertise on the fellowship SOC is a not a barrier (versus a minor or major barrier) 1.83 (1.14–2.98) 
Lack of an adequate core research curriculum is not a barrier (versus a minor or major barrier) 1.65 (1.07–2.56) 

The vast majority of FPDs agreed or strongly agreed that training future researchers is an important component of fellowship (n = 473; 94.8%) and that all fellows in their subspecialty should complete a scholarly project (n = 456; 91.4%). However, 40.5% (n = 202) agreed or strongly agreed that fellowship training should be shortened to 2 years for those interested in clinical or clinician-educator careers. There was significant variability by subspecialty. For example, 64.1% (n = 25) of endocrinology FPDs, 61% (n = 25) of infectious disease FPDs, and 58.6% (n = 34) of emergency medicine FPDs agreed or strongly agreed, whereas the same was true of only 8.3% of cardiology FPDs (n = 3), 11.1% (n = 6) of neonatal-perinatal medicine FPDs, and 18.4% (n = 9) of critical care FPDs. Only 6.6% (n = 33) of FPDs believed fellowship should be shortened to 2 years for all fellows. In multivariable modeling, the odds of wanting to shorten fellowship for certain fellows increased with decreasing program size. Compared with programs with >9 fellows, the OR for programs with 7 to 9, 4 to 6, and 1 to 3 fellows increased from 2.28 (CI: 0.96–5.89) to 5.59 (CI: 2.62–13.41) to 7.98 (CI: 3.78–19.02), respectively. The FPD being not at all or only somewhat satisfied with the overall scholarly activity training experience provided was also significant (OR = 1.65; CI: 1.07–2.55).

We found that pediatric fellows pursue diverse career paths. Positive scholarship experiences during training increase the likelihood that fellows continue to engage in research.15  Thus, well-designed approaches to scholarly activity training can help ensure a future pipeline of pediatric subspecialty investigators, which is critical given their declining numbers.8 

At the same time, although most fellows will not become physician investigators and the majority of programs graduate fellows who pursue careers with focuses in medical education and QI, there was limited variety in the type of scholarly activity proposed to the ABP. This suggests that fellowship programs may be missing important opportunities to help fellows use scholarly projects to explore career interests and develop foundational skills to pursue a niche with greater rigor after fellowship. Nearly 90% of practicing pediatric subspecialists believe that scholarly activity during fellowship should be tailored to fellow career goals.15  Increasing exposure to and mentorship for nonclinical research pursuits, including through dedicated career development sessions, fellow participation in skill-building workshops at regional or national meetings, and identification of departmental, institutional, or even national mentors with training in QI, education, and other disciplines may be helpful.

We also found that the majority of fellows do not publish their fellow project, despite the requirement of producing a work product. Whether publication is the best metric for all fellows is arguable, although the Next Accreditation System has focused on the reporting of traditional productivity metrics.10  It may be that programs are not expecting fellows to publish or that fellows are not successful in publishing. Most FPDs endorse adequate protected time and resources to support fellow scholarly activity, suggesting these are not barriers, at least from the FPD perspective. Our study’s findings are consistent with a survey of practicing pediatric subspecialists, in which 89% felt the time allocated for scholarly activity during fellowship was sufficient.15  In a systematic review, researchers found protected research time was associated with increased publications in medical and surgical specialties.16  However, similar to our study, in a survey of pediatric PDs, researchers found no association between residencies in the top quartile of scholarly productivity and protected research time.12  These findings suggest that protected research time is necessary but insufficient alone to assure productivity.

We found that high productivity was associated with a T32 training grant, FPD publications, SOC faculty expertise, and an adequate core research curriculum. It may be that a program’s research culture leads to selection bias in recruiting fellows more academically inclined or that a strong research culture significantly influences productivity. Regardless, it is logical that local expertise and a strong mentoring team help ensure a well-designed fellow project. If the FPD’s strengths do not lie in research, programs may want to consider selecting a faculty member with research expertise to oversee fellows’ scholarships. Given the influence of the SOC, programs should select SOC faculty with expertise matched to support fellows’ projects. In addition, although most programs reported having statistical support available to fellows, nearly 50% felt it was still a barrier. It may be that statisticians do not prioritize fellow projects, and thus advocating for additional resources specifically for fellows is important. Many FPDs also reported faculty mentorship, particularly within their division, as a barrier. Expanding the available pool of mentors by fostering connections across divisions, departments, and institutions might be necessary to ensure fellows in these programs are not hindered in their career development and scholarship efforts.

A factor we did not assess that may contribute to productivity is the individual fellow’s interest in research and their particular project. A survey of pediatric chief residents found those interested in research were 5 times more likely to be productive.17  Similarly, fellows who lack strong interest in a project may not invest the energy needed to achieve publication. Given that few programs cite projects outside the realm of clinical research, it is worth considering whether assisting fellows to design projects more closely related to their longer-term career goals might influence productivity. In recent years, the ACGME and the ABP have expanded definitions of scholarly activity and allowable work products, yet few trainees pursued projects in these areas. With the most recent changes in ACGME annual reporting requirements, programs must develop a mission and aims.18  We recommend this as an opportunity for programs to explicitly outline scholarship aims, including goals and expectations for productivity, and to carefully consider the types of scholarship opportunities they might offer based on available resources.

The ABP requires that all fellows participate in a core curriculum in scholarly activity and does test key concepts on subspecialty board examinations. However, the ABP’s stated goal for fellow engagement in scholarly activity is acquisition of deeper skills in critical analysis, assimilation of new knowledge into practice, formulation of clear and testable questions, and the advancement of research.14  Yet despite this rationale, >40% of FPDs thought fellowships should be shortened for those with clinically oriented career goals, similar to previous FPD surveys.9  These findings should spur a policy discussion about potential changes to training requirements. This issue is admittedly complex, because in many academic institutions junior faculty must take 100% clinical roles until they are able to acquire funding. Taking time away from research during fellowship training might disadvantage fellows as they try to carve an academic niche or secure an academic position by producing work products during fellowship. Other unintended consequences could include overburdening clinically oriented fellows, the need for programs to recruit additional fellows to meet clinical demands, potentially oversaturating certain fields, and increased clinical demands for faculty, hindering their productivity.

At the same time, physician-scientist training programs, mentored opportunities for junior faculty (such as T32s or KL2s), or a postfellowship research year may be required to support more rigorous research training for the small group interested in becoming independent investigators. Although funding to support an additional fellowship year is often limited, shortening fellowship for some fellows could allow resources to be diverted from fellows who only want a shorter experience to fellows with long-term potential for research.

Our study has several limitations. First, because this was a cross-sectional survey, we are only able to determine associations, not causality. Second, there may be response bias from FPDs more interested in scholarly activity training, although we did have a high response rate. Third, FPDs may be unaware of fellow publications accepted after fellowship, leading to undercounting. Fourth, for many programs, the presence of an academically productive FPD, research-oriented faculty, or training grants likely influences the candidates they recruit, influencing productivity associations. We also did not ask FPDs if perceived lack of research interest from fellows was a barrier. Finally, this survey does not reflect fellow perspectives, which will be critical to understanding their desires for training or perceived barriers to publishing their scholarly work.

We found that FPDs report graduates pursue a variety of career paths after fellowship, but the types of scholarly activities proposed to the ABP for graduates is limited. Most FPDs report that fellows are not publishing, despite a median of 17 months of protected research time. FPD productivity, SOC expertise, and division-specific research curriculum are associated with fellow publications. Ensuring a culture of research within a fellowship program, including a program director (or designated research director) who publishes frequently and a SOC with project-specific expertise, may provide the mentorship needed to take projects to publication. We would encourage greater alignment between scholarly projects and fellow career goals and encourage FPDs to clearly define their program mission and aims. The fellowship community may consider reevaluating the duration of the fellowship for some fellows, particularly for those pursing a clinical or clinician-educator career.

We thank the following SPIN Steering Committee members who collaborated on this project and were instrumental in recruiting participants: Richard Mink (Director), Bruce Herman, Alan Schwartz, Carol Carraccio, Sarah Pitts, Brad Robinson, Shubhika Srivastava, Mary Moffatt, David Turner, Angela Czaja, Jill Fussell, Pam High, Deborah Hsu, Melissa Langhan, Diane Stafford, Tandy Aye, Cary Sauer, Jennifer Kesselheim, Mark Atlas, Angie Myers, Kammy McGann, Christiane Dammann, Patricia Chess, John Mahan, Susan Halbach, and Megan Curran.

We thank SPIN, and especially Beth King, for support in facilitating this project. We thank the ACGME for providing us with data us data on number of fellows by fellowship program in 2018–2019. We also thank all the FPDs who participated by completing the survey. We presented this work as a platform presentation at the Association of Pediatric Program Directors annual spring meeting in June 2020.

Dr Abramson conceptualized and designed the study, designed the data collection instrument, drafted the initial manuscript, and reviewed and revised the manuscript; Drs Weiss, Naifeh, Stevenson, Duncan, and Li helped with critical review and the drafting of the data collection instrument, drafting of the manuscript, and review and revision of the manuscript; Ms Mauer and Dr Gerber conducted data analyses and participated in review and revision of the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

FUNDING: No external funding.

COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/2020-025296.

ABP

American Board of Pediatrics

ACGME

Accreditation Council for Graduate Medical Education

CI

confidence interval

FPD

fellowship program director

OR

odds ratio

QI

quality improvement

SOC

Scholarship Oversight Committee

SPIN

Subspecialty Pediatric Investigator Network

1
American Board of Pediatrics
.
Entrustable Professional Activities for subspecialists
. Available at: https://www.abp.org/subspecialty-epas. Accessed January 28, 2020
2
Mink
RB
,
Myers
AL
,
Turner
DA
,
Carraccio
CL
.
Competencies, milestones, and a level of supervision scale for Entrustable Professional Activities for scholarship
.
Acad Med
.
2018
;
93
(
11
):
1668
1672
3
Committee on Pediatric Research
.
Promoting education, mentorship, and support for pediatric research
.
Pediatrics
.
2014
;
133
(
5
):
943
949
4
The American Board of Pediatrics
.
Scholarly activity
. Available at: abp.org/content/scholarly-activity. Accessed January 27, 2020
5
Epstein
RM
.
Assessment in medical education
.
N Engl J Med
.
2007
;
356
(
4
):
387
396
6
Shurin
SB
.
The American Pediatric Society commentary: the future of pediatric physician scientists is in our hands
.
Arch Pediatr Adolesc Med
.
2012
;
166
(
6
):
503
505
7
Weiss
P
,
Myers
AL
,
McGann
KA
, et al
.
Funding sources and perceived financial insecurity in pediatric subspecialty fellowship programs
.
Acad Pediatr
.
2019
;
19
(
7
):
815
821
8
Macy
ML
,
Van
KD
,
Leslie
LK
,
Freed
GL
.
Engagement in research among pediatric subspecialists at the time of enrollment in maintenance of certification, 2009-2016
.
Pediatr Res
.
2020
;
87
(
6
):
1128
1134
9
Freed
GL
,
Dunham
KM
,
Moran
LM
,
Spera
L
,
McGuinness
GA
,
Stevenson
DK
.
Fellowship program directors perspectives on fellowship training
.
Pediatrics
.
2014
;
133
(
suppl 2
):
S64
S69
10
Philibert
I
,
Lieh-Lai
M
,
Miller
R
,
Potts
JR
 III
,
Brigham
T
,
Nasca
TJ
.
Scholarly activity in the next accreditation system: moving from structure and process to outcomes
.
J Grad Med Educ
.
2013
;
5
(
4
):
714
717
11
Mink
R
,
Schwartz
A
,
Carraccio
C
, et al.;
Steering Committee of the Subspecialty Pediatrics Investigator Network
.
Creating the subspecialty pediatrics investigator network. [published correction appears in J Pediatr. 2019;207:269]
.
J Pediatr
.
2018
;
192
:
3
4.e2
12
Abramson
EL
,
Naifeh
MM
,
Stevenson
MD
, et al
.
Research training among pediatric residency programs: a national assessment
.
Acad Med
.
2014
;
89
(
12
):
1674
1680
13
Abramson
EL
,
Naifeh
MM
,
Stevenson
MD
, et al
.
Scholarly activity training during residency: are we hitting the mark? A national assessment of pediatric residents
.
Acad Pediatr
.
2018
;
18
(
5
):
542
549
14
Hijazi
M
,
Downing
SM
.
Objective structured clinical examinations as an assessment method in residency training: practical considerations
.
Ann Saudi Med
.
2008
;
28
(
3
):
192
199
15
Freed
GL
,
Dunham
KM
,
Moran
LM
,
Spera
L
,
McGuinness
GA
,
Stevenson
DK
.
Pediatric subspecialty fellowship clinical training project: recent graduates and midcareer survey comparison
.
Pediatrics
.
2014
;
133
(
suppl 2
):
S70
S75
16
Stevenson
MD
,
Smigielski
EM
,
Naifeh
MM
,
Abramson
EL
,
Todd
C
,
Li
S-T
.
Increasing scholarly activity productivity during residency: a systematic review
.
Acad Med
.
2017
;
92
(
2
):
250
266
17
McHenry
MS
,
Abramson
EL
,
McKenna
MP
,
Li
S-T
.
Research in pediatric residency: national experience of pediatric chief residents
.
Acad Pediatr
.
2017
;
17
(
2
):
144
148
18
Accreditation Council for Graduate Medical Education
.
ACGME common program requirements (fellowship)
. Available at: https://www.acgme.org/Portals/0/PFAssets/ProgramRequirements/CPRFellowship2020.pdf. Accessed May 20, 2020

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data