Trends in Race/Ethnicity of Pediatric Residents and Fellows: 2007–2019

We performed a repeated cross-sectional study of race and ethnicity in pediatric residents and subspecialty fellows using publicly available National Graduate Medical Education (GME) Census reports from 2007, when race and ethnicity data were first included, to 2019. We used the AAMC Medical Minority Applicant Registry definition of URiM: self-identification as AA or Black, Hispanic or Latino, AI or AN, or NH or PI.²⁹  This report follows the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline for cross-sectional studies. This study was considered exempt from human subjects research by the Wake Forest University Health Sciences Institutional Review Board.

We manually extracted self-reported race and ethnicity data from 2007 to 2019 for US pediatric residents and fellows from the 17 pediatric subspecialties that were available from GME Census reports. These National GME Census data are reported annually in the Journal of the American Medical Association, which confirms the data jointly collected by the AAMC and American Medical Association via surveys to directors of programs accredited by the ACGME.^30–42  Confirmation of status for all active physicians-in-training ranged from 92.5% in 2010 to 97.3% in 2019. Race and ethnicity data from specific institutions were not available. Of note, during the study period, from 2007 to 2014, the National GME Census asked for race and Hispanic ethnicity in 2 separate questions; after 2014, the National GME Census imported self-designated race and ethnicity from AAMC databases, which included a “multiracial” category, referring to trainees who self-identified as >1 race. Throughout all data source years, a person of Hispanic ethnicity could be of any race. We excluded combined residencies not restricted to pediatric trainees and subspecialties in which a completion of a pediatric residency was not required, such as the pediatric surgical subspecialties. We obtained 2019 US population data from the US Census Bureau.⁴³  We obtained yearly population distribution of race and ethnicity estimates on the basis of analysis of the American Community Survey, 1-year estimates from 2008 to 2019.⁴⁴  The racial and ethnic categories used throughout this article reflect those reported in the original data sources.

Results were analyzed by using descriptive statistics. Inclusion of racial and ethnic groups into the URiM category was based on the AAMC definition of URiM. Therefore, the unknown and multiracial race and ethnicity categories were not included in the URiM category. Proportions (URiM out of total) were calculated for each year and specialty; the Clopper-Pearson exact binomial method was used to calculate 95% confidence intervals (CIs). To determine if there was a significant trend in URiM representation over time, proportions (URiM out of total) were compared over time to the Cochran-Armitage test from all years (2007–2019) for pediatric residents and fellows from 15 of the 17 available pediatric subspecialties. The clinical informatics subspecialty, which began reporting fellows in 2015, did not report any URiM fellows over the study period; the transplant hepatology subspecialty, which began reporting fellows in 2009, only reported 7 total URiM fellows; these 2 subspecialties were included in the overall analyses, but URiM trends within the subspecialties were not included in the analysis, given the small sample size. The child abuse subspecialty began reporting fellows in 2012. Results were considered statistically significant at a 2-tailed P value < .05. SAS (version 9.4; SAS Institute, Inc, Cary, NC) was used for all analyses.

A total of 109 344 pediatric residents and 44 420 subspecialty fellows were included in the sample. Overall, the unknown race and ethnicity category accounted for 10% of pediatric residents and 5.2% of subspecialty fellows; the multiracial category (available from 2015 to 2019) accounted for 1.2% of pediatric residents and 4% of subspecialty fellows, or 2% of all trainees.

Between 2007 and 2019, the annual number of pediatric trainees increased from 7964 to 8950 residents and 2684 to 3966 subspecialty fellows; the annual number of URiM pediatric trainees also increased over time, from 1277 to 1478 residents and 382 to 532 subspecialty fellows (Table 1). However, over time, the trend in proportions of URiM trainees was unchanged in pediatric residencies (16% in 2007 [CI: 15.2–16.9] to 16.5% in 2019 [CI: 15.8–17.3]; P = .98 for trend from 2007 to 2019) and, overall, decreased for subspecialty fellows (14.2% in 2007 [CI: 12.9–15.6] to 13.5% in 2019 [CI: 12.4–14.6]; P = .002 for trend from 2007 to 2019). Significantly decreased trends over time in URiM fellow representation also occurred for neonatal-perinatal medicine (P < .001), infectious diseases, (P < .001), and critical care (P = .006). Significantly increased trends in URiM fellow representation existed in endocrinology (P = .002) and pulmonology (P = .009).

In 2019, the percentage of URiM pediatric trainee representation was considerably lower in comparison with those groups’ representation in the US population: AA or Black, 5.6% vs 13.4%; AI or AN, 0.2% vs 1.3%; NH or PI, 0.1% vs 0.2%; Hispanic or Latino, 9.7% vs 18.3%; multiracial, 2% vs 2.8% (Fig 1).⁴³  Over time, the proportion of URiM pediatric trainee representation remained considerably lower in comparison with those groups’ representation in the US population (Fig 2).⁴⁴ 

This repeated cross-sectional study of pediatric residents and subspecialty fellows from 2007 to 2019 is one of the first to report trends in URiM pediatric trainee representation. Overall, URiM representation in pediatric training programs remains low. For example, we found that the proportion of URiM representation in residents remained stagnant and significantly declined overall for subspecialty fellows; only the endocrinology and pulmonology subspecialties revealed significant improvement in URiM subspecialty fellow representation over time. In 2019, URiM representation among pediatric trainees did not reflect the racial and ethnic diversity of the US population, which was a consistent pattern over time.

Although further research is needed to delineate the reasons for the striking lack of URiM representation in pediatrics in general and over time, several possibilities exist that span the entire educational continuum. Previous studies have revealed that URiM medical school applicants and matriculants have lower representation compared with the US population,²⁴  indicating that the pattern of underrepresentation in medicine develops before medical school application. Contributing factors to low proportions of URiM medical school applicants include educational disparities starting in primary and secondary education, such as underinvestment in public schools,⁴⁵  educational resource allocation disparities,⁴⁶  and policies promoting racially segregated schools, including “school choice” and linking enrollment with housing segregation patterns.⁴⁷ 

Beyond primary and secondary education, the lack of diversity spans the processes of recruitment, inclusion, and retention, including experiences in the learning and clinical environment, all of which are permeated by racism and bias. For example, in undergraduate education, anti–affirmative action policies have led to sharp declines in the enrollment of underrepresented minorities.^48,49  Public colleges and universities remain inaccessible to Black and Hispanic students,⁵⁰  and the percentage of Black men enrolled in medical school decreased between 1978 and 2014 and, since then, has only slightly increased.^51,52  Furthermore, disparities in the proportion of URiM medical school matriculants versus applicants, particularly for Black students, may be due to implicit bias and racism in the admissions process.⁵³ 

Once in medical school, URiM medical students encounter differential access to opportunities, receive lower clinical grades and standardized test scores and fewer inductions into Alpha Omega Alpha honor society, and experience racism and implicit bias in the learning and clinical environments, which lead to higher attrition rates and affect competitiveness as a residency applicant and pursuit of academic careers.^54–58  These studies highlight that educational disparities for URiM medical students exist even before entering the field of pediatrics. Efforts to increase diversity among pediatric trainees are limited by the amount of URiM graduating medical students, highlighting the need to develop longitudinal diversity initiatives, including mentorship, in the early developmental and educational stages.⁵⁹ 

In pediatrics, several factors contribute to the lack of diversity in residency, subspecialty fellowship, and academic medicine. For pediatric residencies, URiM medical students report that mentorship and URiM pediatric representation positively influence the choice to pursue academic pediatrics.⁶⁰  Additionally, negative experiences in the learning and clinical environment for residents may lead to higher turnover and job satisfaction.^61,62  Studies from other specialty residency programs reveal that URiM residents have higher attrition rates.^63,64  Financial considerations are frequently cited as a factor when residents consider pediatric subspecialties, with higher debt burdens associated with a push toward general practice.^65,66  It has been revealed that anticipated educational debt can pose a disproportionate barrier for individuals from socioeconomically disadvantaged backgrounds and/or URiM groups,^13,67  likely influencing subspecialty choice and a potential career in academic medicine.^60,68 

Among academic pediatricians, implicit bias may negatively affect minority recruitment efforts from residents to faculty members^68,69  in addition to lack of an explicit URiM recruitment priority by programs.⁷⁰  The role of faculty mentors strongly influences choosing a career in academic pediatrics,^65,71  particularly for URiM pediatricians.^68,69  In our study, only the endocrinology and pulmonology subspecialty fellowships significantly increased URiM representation over time, which may reflect mentorship by URiM subspecialists, although further research is necessary to determine race and ethnicity representation of subspecialty pediatricians. Additionally, the endocrinology and pulmonology fellowships typically have higher percentages of unfilled positions, which may be more likely filled with diverse international medical graduates.

In reference to retention, URiM faculty report less satisfaction with their careers and are more likely to leave academic medicine,⁷²  creating a vacuum of representation for URiM trainees, including in leadership positions.⁶⁸  Given that academic pediatric faculty are involved in career mentoring, advising, and the trainee recruitment process, the racism and implicit bias experienced by URiM faculty in academic medicine may contribute to differential access to URiM mentorship and deter URiM trainees from considering careers in pediatric subspecialties and academic medicine. Lastly, the disparity in numbers of trainees and faculty members from URiM groups is exacerbated among those with intersectional identities. Intersectionality is often defined as the way social categorizations, including race, ethnicity, class, and gender identity, create overlapping and interdependent systems of discrimination.⁷³  The majority of practicing pediatricians in the United States identify as women; the AAMC reports that 71% of pediatric residents were women in 2020.⁷⁴  In 2018–2019, 41% of full-time faculty identified as women, yet only 13% of those were from a URiM group.⁷⁴  Women with intersectional identities are known to experience bias and discrimination at higher rates than other groups,⁷⁵  possibly contributing to attrition and lack of representation of URiM women in academic medicine.^76,77 

The AAMC has published a diversity and inclusion strategic planning guide providing a useful framework to improve the diversity of training programs.⁷⁸  Furthermore, in 2019, the ACGME instituted a new common program requirement on diversity:

A similar requirement was created by the Liaison Committee on Medical Education in 2009 for medical schools and was last updated in 2015.¹⁸  For recruitment, residency and fellowship training programs should holistically identify potential interviewees and use enrollment management recognizing the success of such programs.^20,21  Additionally, programs should standardize the interview process and highlight diversity efforts.²⁰  Barriers should be removed for disadvantaged URiM applicants applying to residency programs, such as application costs and/or scheduling or attending interviews, given the disproportionate debt incurred by URiM applicants.^66,79,80  Pediatric programs should provide ongoing implicit bias training for members of recruitment committees.⁸¹  Furthermore, pediatric training programs can work to mitigate disparities in URiM representation by establishing mentorship programs along the educational spectrum, particularly for medical students.⁶⁰  Regular advising meetings with trainees should include discussions on residency, fellowship, and academic faculty preparation and identify gaps in clinical exposure, scholarship, and mentorship. Retention efforts should include examination of the training environment pertaining to diversity, equity, and inclusion. There should be an annual examination within pediatric programs of recruitment and retention rates of trainees and faculty members, such as a diversity dashboard^82,83 ; withdrawals and dismissals among URiM trainees should be analyzed comprehensively. Lastly, professional pediatric societies should also examine the racial and ethnic composition of their members and leadership and take steps to improve recruitment and retention of URiM pediatricians. Beyond pediatrics, rigorous collection of demographic data at a granular level and alignment of this collection across societies and medical institutions may be used to ascertain trends more accurately and assess whether interventions have been successful.

There are some limitations to this study that should acknowledged. First, National GME Census data are imported from the AAMC’s databases, and, when self-reported race and ethnicity were not available, they were reported by the program directors. Second, the change in data collection for the multiracial category may have biased our race and ethnicity results toward observing underrepresentation in URiM groups; however, only 1.2% of residents and 4% of specialty fellows identified as multiracial from 2015 to 2019, and the category may have included individuals who were not URiM.

Pediatric residents and subspecialty fellows are the future pediatric workforce. In >10 years, there has been little improvement in the representation of URiM trainees. There is a critical need to recruit and retain URiM pediatric residents and fellows in all subspecialties to reflect the increasingly diverse populations we serve and to mitigate health inequities. Pediatric residency programs can help alleviate disparities in fellowship programs through early mentorship. Fellowship programs should holistically identify applicants recognizing that higher barriers exist for URiM trainees. Subspecialty societies should identify future subspecialists by creating longitudinal mentorship programs for underrepresented undergraduate and medical students, such as those developed by the Academic Pediatrics Association.^84,85  Professional societies should critically examine the racial and ethnic makeup of their members and leadership. Retention efforts should include an examination of the training environment as it relates to diversity, equity, and inclusion, and dashboards should be created to monitor progress. Standardizing the collection of granular demographic data across societies and institutions may help track recruitment and retention trends, in addition to developing and determining the success of interventions.

AA

African American

AAMC

Association of American Medical Colleges

ACGME

Accreditation Council for Graduate Medical Education

AI

American Indian

AN

Alaskan native

CI

confidence interval

GME

graduate medical education

NH

native Hawaiian

PI

Pacific Islander

URiM

underrepresented in medicine