Introduction

Over the past decade, studies have demonstrated that diversity as a core value can significantly drive and improve innovation [1]. Specific to medicine, recent studies have shown convincing evidence of the necessity for a diverse health care workforce to deliver the high-value medical education, advancements in research, and health care access to under-represented minorities (URMs) [2, 3]. For example, URM physicians are more likely to work in URM communities [4]. Further, URM patients perceive higher contentment with medical care received when the care is provided by URM physicians [3].

While the importance of a diverse medical workforce has been established, has medicine met this need? An examination of diversity within graduate medical education might be a potential means of assessing future diversity in medicine. The Council on Graduate Medical Education (COGME) was established in 1986 by Congress to provide an on-going assessment of physician workforce trends and training issues. In 1993 and again in 2005, the COGME issued reports specifically examining the advancement of goals to increase under-represented minorities (URMs) in the medical profession. Specifically, the COGME reports emphasized that “increasing the number of URM physicians is an important step for improving health care for minorities and underserved populations” [5, 6].

Partly as a result of these reports, ethnic and gender diversity in medicine has become a subject of considerable attention within several medical specialties [7–9]. Recently, there have been a number of publications examining ethnic and gender diversity in the field of radiology. In 2014, Lightfoote et al. published a two-part article, “Improving Diversity, Inclusion, and Representation in Radiology and Radiation Oncology…,” which, on behalf of the American College of Radiology, examined ways to improve diversity and inclusion within the radiology medical specialty [10, 11].

Among the recommendations of the COGME reports was that “more research is needed to evaluate obstacles or motivations for minority entry into primary care or specialty residency programs.” As a specialty, radiology has begun this process. In 2014, Chapman et al. published a seminal article titled “Current Status of Diversity by Race, Hispanic Ethnicity, and Sex” in Radiology. Using data available from the American Medical Association (AMA), this article demonstrated a statistically significant difference between the percentage of women and under-represented minorities in the US census population and all levels of training in radiology [4]. To date, however, no one has examined diversity in radiology fellowships. The purpose of this article is to examine gender and URM diversity trends in all (Accreditation Council for Graduate Medical Education (ACGME)-certified radiology fellowships.

Material and Methods

Institutional Review Board (IRB) evaluation and exemption were granted for the study, as primary data were obtained from publicly available registry sources, with no identifiable private or protected information. Each year, the American Medical Association (AMA) and the Association of American Medical Colleges (AAMC) administer the National Graduate Medical Education Census. Directors of residency and fellowship programs that are accredited by the Accreditation Council for Graduate Medical Education (ACGME) are asked to complete the program survey component of the census. The data from these surveys is published each year in the Journal of the American Medical Association (JAMA). For this study, ethnic and gender characteristics of participants in ACGME certified radiology fellowships (as defined in Table 1) and all medical fellowships were obtained from JAMA publications containing the National GME census data from 2005 to 2012 [12–19]. Further, ethnic and gender characteristics of radiology residents and medical students from 2005 to 2012 were obtained from the GME census and from the AAMC, respectively [7, 20–26]. Finally, US ethnic and gender population statistics were obtained from the 2010 US census data [27].

Table 1 ACGME-certified radiology fellowships
Full size table

According to the US Department of Health and Human Services, under-represented minorities (URMs) are defined as “racial and ethnic populations who are underrepresented in a designated health profession discipline relative to the percentage of that racial or ethnic group in the total population” [28]. Ethnicity is defined by the categories within the GME survey (National Center for Education Statistics 1997). Black or African American is defined as a person having origins in any of the black racial groups of Africa. Hispanic or Latino is defined as a person of Cuban, Mexican, Puerto Rican, South or Central American, or other Spanish culture or origin, regardless of race. Based on these categories and definitions of ethnicity and previously mentioned definition of URM, Figs. 1, 2, and 3 summarize gender and ethnic fellowship participant demographic percentage distribution for each ACGME certified radiology fellowship as defined above, from 2005 to 2012. Figures 4, 5, and 6 summarize the percentages of females, African-Americans, and Latinos in all radiology fellowships (combined), medical fellowships, radiology residencies, medical residencies, medical school graduates, and US census, respectively.

Fig. 1
figure 1

Bar graph comparing the percentages of women in each radiology fellowship type from 2006 to 2013

Full size image
Fig. 2
figure 2

Bar graph comparing the percentages of African-Americans in each radiology fellowship type from 2006 to 2013

Full size image
Fig. 3
figure 3

Bar graph comparing the percentages of Latinos in each radiology fellowship from 2006 to 2013

Full size image
Fig. 4
figure 4

Bar graph comparing the percentages of women at progressive levels of medical education and from 2006 to 2013 and the US census (2010)

Full size image
Fig. 5
figure 5

Bar graph comparing the percentages of African Americans at progressive levels of medical education from 2006 to 2013 and the US census (2010)

Full size image
Fig. 6
figure 6

Bar graph comparing the percentages of Latinos at progressive levels of medical education from 2006 to 2013 and the US census (2010)

Full size image

Using the data contained in these figures, Fisher’s exact test and chi-squared test were used to determine statistically significant differences (p < .05) between the percentage ethnic and gender distributions within each radiology fellowship and

  1. 1.

    All fellows in ACGME certified radiology fellowships.

  2. 2.

    All radiology residents in ACGME certified radiology residencies.

  3. 3.

    All medical fellows in ACGME certified fellowships, regardless of specialty.

  4. 4.

    All medical residents in ACGME certified residencies, regardless of specialty.

  5. 5.

    All graduating medical students

  6. 6.

    US census.

Results

The results of statistical analyses are summarized and displayed for each radiology fellowship in Tables 2, 3, 4, 5, 6, 7, and 8.

Table 2 Comparative statistical analysis of percentage female and URM abdominal radiology fellows. Data that demonstrated significant differences between each radiology fellowship and comparison groups were in bold
Full size table
Table 3 Comparative statistical analysis of percentage female and URM endovascular neouroradiology radiology fellows. Data that demonstrated significant differences between each radiology fellowship and comparison groups were in bold
Full size table
Table 4 Comparative statistical analysis of percentage female and URM interventional radiology fellows. Data that demonstrated significant differences between each radiology fellowship and comparison groups were in bold
Full size table
Table 5 Comparative statistical analysis of percentage female and URM musculoskeletal radiology fellows. Data that demonstrated significant differences between each radiology fellowship and comparison groups were in bold
Full size table
Table 6 Comparative statistical analysis of percentage female and URM neuroradiology radiology fellows. Data that demonstrated significant differences between each radiology fellowship and comparison groups were in bold
Full size table
Table 7 Comparative statistical analysis of percentage female and URM nuclear medicine fellows. Data that demonstrated significant differences between each radiology fellowship and comparison groups were in bold
Full size table
Table 8 Comparative statistical analysis of percentage female and URM pediatric radiology fellows. Data that demonstrated significant differences between each radiology fellowship and comparison groups were in bold
Full size table

Abdominal Radiology

Abdominal radiology demonstrated sporadically lower percentages of women in 2006, 2009, and 2013, particularly when compared to percentage of female medical school graduates. Additionally, in 2013, the percentage of women in abdominal radiology fellowship was significantly lower than the average of all radiology fellows, which had not been seen in any previous years. For African-American URMs, the percentage of abdominal fellows was significantly lower when compared to the US census in 2006, 2009, 2010, and 2013. Latino URMs demonstrated the least amount of statistically significant variations in percentage of fellows, with statistically lower percentages when compared to the US census in 2006 and 2011.

Endovascular Neuroradiology

Of all of the radiology fellowship, endovascular neuroradiology had the fewest overall fellows each year. Because of this, statistical significance was rarely reached. However, in spite of this, the percentage of women endovascular neuroradiology fellows was significantly lower than the census from 2009 to 2011 and significantly lower than the percentage of female medical residents in 2011. There was no statistically significant difference in the percentage of Latino or African-American URMs in endovascular neuroradiology and any of the subcategories studied.

Interventional Radiology

Of all of the ACGME-certified radiology fellowships, interventional radiology has the largest number of programs. However, interventional radiology demonstrates the least amount of diversity, as measured by this study. At every level of medical education, the percentage of female interventional radiology fellows was comparatively significantly lower, regardless of year. From 2006 to 2008, the percentage of African American interventional fellows only demonstrated a significantly lower percentage, when compared to the US census. However, from 2009 to 2013, the percentages of African-American interventional radiology fellows were significantly lower when compared to all radiology residents and all radiology fellows. The percentages of Latino interventional fellows generally compared favorably. With the notable exception of 2009, the percentage of Latino fellows was only significantly lower than the US census.

Musculoskeletal Radiology

From 2006 to 2007, there were no statistically significant differences in percentages of female or URM musculoskeletal radiology fellows. However, from 2008 to 2012, the percentage of female musculoskeletal radiology fellows was significantly lower than the US census, medical school graduates, and medical school residents. Further, in 2008 and 2009, female musculoskeletal radiology fellows demonstrated a significantly lower percentage when compared to all medical fellows. African-American and Latino URMs demonstrated significantly lower percentages when compared to the 2012 and 2013 US census, respectively.

Neuroradiology

Of all radiology fellowships, neuroradiology has the most fellows. However, comparatively, the percentages of females in neuroradiology are lower than the percentages of women in almost all of the compared levels of medical education, with the notable exception of all radiology fellows. Similarly, the percentages African-American neuroradiology fellows demonstrated significantly lower percentages, when compared to almost all levels of medical education, again, with the notable exception of all radiology fellows. With respect to percentages of neuroradiology fellows, Latinos demonstrated the fewest significant differences across most medical education categories. However, the percentages of Latino neuroradiology fellows are significantly lower than the US census, across all years.

Nuclear Medicine

Similar to endovascular neuroradiology, the overall numbers of fellows in nuclear medicine are low, such that reaching statistical was difficult. Within this limitation, the only statistically significant percentage differences were seen in 2013, where the percentage of female fellows in nuclear radiology was lower than the US census and medical school graduates.

Pediatric Radiology

In terms of percentages of female fellows, pediatric radiology was the only fellowship subspecialty of radiology to trend differently from the other fellowships. Across almost all years examined, the percentages of female pediatric radiology fellows were significantly higher than all radiology fellows, all medical fellows, and all radiology residents. However, the percentages African-American pediatric radiology fellows demonstrated significantly lower percentages, when compared to the US census. Additionally, in 2009 and 2011, the percentages of African American pediatric radiology fellows were significantly lower than all radiology fellowships. The percentages of Latino pediatric radiology fellows did not demonstrate any significant differences when compared to any of the medical education levels examined.

Discussion

In 2015, Deville et al. published a study examining gender and ethnicity among all medical specialties. They found that radiology, orthopedics, and otolaryngology demonstrated statistically significant lower numbers of women and URMs, illustrating the diversity problem in radiology [29]. In the second part of their two-part article, “Improving Diversity, Inclusion, and Representation in Radiology and Radiation Oncology,”, Lightfoote et al. discussed the development and implementation of concrete strategies designed to eliminate the current subspecialty disparity [11]. As a part of that discussion, this study examined the percentages of females and under-represented minorities within ACGME certified fellowships.

There are currently seven different ACGME-certified fellowships, ranging from 1 to 3 years in duration. From this study, there is clear evidence of gender disparities within the radiology fellowships. Almost all fellowships demonstrated some levels of gender disparity, with interventional radiology and neuroradiology demonstrating the highest levels of disparity. Pediatric radiology, on the other hand, demonstrated higher than expected percentages of female fellows. In the analysis of the significant differences between the percentages of female fellows and other levels of medical education, a pattern of significantly lower percentage female radiology fellows, compared to all medical residents and fellows, seems to emerge. This may indicate that prospective female medical graduates are choosing specialties other than radiology, limiting the pool of female radiology residents to enter fellowships. Figure 4 further demonstrates the precipitous drop in the percentage of females from medical school to radiology residency. Hewett et al. examined the application processes in one radiology residency program to determine if there was any gender bias in the selection of radiology residents. They found that the pipeline of female medical students pursuing a career in radiology appeared to be the limiting factor rather than bias [30].

African-American radiology fellows demonstrated a similar, albeit less pronounced, pattern of disparity to what was seen in gender. In interventional radiology and neuroradiology fellowships, a pattern of lower percentage African-American radiology residents and fellows, when compared to all medical residents and fellows may again demonstrate the need to recruit more African-American medical school graduates into radiology.

Latino radiology fellows demonstrated the least amount of disparity. Particularly within the larger fellowships (interventional radiology and neuroradiology), significant disparities in the percentage of Latino fellows were mostly seen when compared to the US Census.

In 2015, Phitayakorn et al. demonstrated “that a program’s existing diversity and climate for minorities is weighted more heavily by minority candidates” [31]. The lack of minorities in specialties such as radiology can have a significant effect on the recruitment of other minorities and, thus, have a downstream effect on diversity in fellowships. This illustrates the importance of pipeline initiatives and recruitment of women and URMs into radiology to address the lack of diversity.

Limitations

There are several limitations to the study that should be noted. First, this study only included ACGME-certified fellowships. However, fellowships such as breast imaging are not included. It is of particular significance since as many as 5 % of radiology residents were interested in breast imaging, which was seventh on a choice list of fellowships [32]. Secondly, there is no clear definition of “other” as an ethnic category. For this reason, this category was omitted from this study.

In summary, this study demonstrates high levels of gender and ethnic disparities in radiology fellowships. This study is the first step toward addressing this issue. Potentially, these disparities could be addressed through more aggressive outreach, pipeline programs, and mentoring. Further studies on workforce diversity beyond post-graduate training are also needed. Databases that track gender and ethnic data are needed to further advance the goals of eliminating disparities within the radiology profession.