COVID-19 Research Delays Disproportionately Affected Pediatrician-Scientists from Backgrounds Underrepresented in Medicine

The COVID-19 pandemic dramatically slowed biomedical research. Supply chain disruptions and social distancing hindered laboratory work; hiring and funding restrictions reduced staff; and school and childcare center closures increased non-work demands on many researchers.[1,2] COVID-19-related challenges doubly affected physician-scientists: unprecedented decreases in elective medical care reduced the clinical revenues which support academic healthcare centers’ funding, and many physician-scientists had to pause their research and prioritize clinical work.[3]

Even before the COVID-19 pandemic, individuals of minority racial and ethnic backgrounds, individuals with disabilities, and socioeconomically disadvantaged individuals, collectively referred to as backgrounds underrepresented in medicine (URIM), were underrepresented among physician-scientists.[4,5] URIM individuals and women account for a smaller percentage of MD-PhD students than of the population at large [6]; women and URIM MD-PhD students have lower graduation rates than their peers,[6] and fewer of them achieve independently-funded research careers.[7]

Considering the historical underrepresentation of these groups among researchers and the pandemic’s negative impact on science, several studies have shown that female researchers were more affected by pandemic-related delays than their male colleagues. Across disciplines including medicine, life sciences, physical sciences, and economics, papers published by women fell significantly more than papers published by men during the pandemic.[8,9] Pandemic lockdown measures disproportionately affected women’s careers [10]; women bore the brunt of research disruptions,[11] and women in heterosexual couples with children decreased their work hours more than men.[12]

Compared with the disproportionate effect of the COVID-19 pandemic on women, little has been published about the experience of other historically disempowered groups. One study of the impacts from the pandemic on pediatrics faculty suggested that URIM individuals were also disproportionately affected, but that result was not statistically significant.[13] Articles have commented that the COVID-19 pandemic worsened the challenges faced by URIM physician-scientists,[7] but those disparities have not been quantified.

In this study, we surveyed early-career pediatrician-scientists — with a particular focus on the experiences of individuals from URIM backgrounds — to explore how the COVID-19 pandemic affected their scholarly progress. Considering the barriers faced by URIM physician-scientists and the possibly disproportionate effects of the pandemic on those individuals, we sought to highlight the ways in which URIM individuals may need additional support in order to foster a diverse future workforce in academic medicine.[5]

Methods

To explore the effects of the COVID-19 pandemic on early-career pediatrician scientists, we surveyed current scholars and recent alumni (≤5 years from completion) from the Pediatric Scientist Development Program (PSDP): an intensive NIH/NICHD K12- and foundation-funded career development program focused on fostering the next generation of diverse pediatrician-scientists through a multilayered mentorship model. PSDP participants are selected with an eye towards diversity, including gender, race and ethnicity, and subspecialty. The PSDP leadership wrote survey questions addressing participants’ demographics, current and anticipated academic progress, and career plans. (Table I). Four times during the pandemic (May 2020, August 2020, November 2020, and November 2021), we sent emails to the entire PSDP email list with invitations to participate in the study and links to an anonymous Qualtrics survey form with our questions. The survey was completely anonymous to encourage honest responses to sensitive questions and to facilitate rapid institutional review board approval, allowing us to disseminate the surveys while the effects of the pandemic were still ongoing. This study was approved as exempt by the Weill Cornell Medicine Institutional Review Board (protocol number 21-08023818).

Table I.

Survey questions used to explore the effects of the COVID-19 pandemic on the research progress of early-career pediatrician-scientists participating in the Pediatric Scientist Development Program (PSDP).

Which year did you enter the PSDP? (Year)

What gender do you identify as? (Female, Male, Other)

What race do you identify as? (Asian, Black or African American, White, More than one race)

What ethnicity do you identify as? (Hispanic or Latino, Not Hispanic or Latino)

Do you have a disability? (NIH definition: a person with a physical or mental impairment that substantially limits one or more major life activities) (Yes, No)

Do you come from a disadvantaged background? (NIH definition: from a family with an annual income below established low-income thresholds and/or from an educational environment such as that found in certain rural or inner-city environments that has demonstrably and directly inhibited you from obtaining the knowledge, skills, and abilities necessary to develop and participate in a research career) (Yes, No)

My research productivity will be negatively impacted in the short-term (~3 months) (Strongly Agree…Strongly Disagree)

My research productivity will be negatively impacted in the long-term (>3 months) (Strongly Agree…Strongly Disagree)

Has the COVID-19 pandemic caused you to reconsider your career as a physician scientist? (Yes, No)

I have changed my intended career path/specialty as a result of COVID-19 (Strongly Agree…Strongly Disagree)

Is there anything else you would like to add regarding the impact of COVID-19 on your career? (Free-text)

This a subset of the questions from the longer survey; the remaining questions were not used in this study. The full survey is available on request from the authors.

We used descriptive statistics to summarize survey responses. Many survey questions used 5-point Likert scales ranging from “strongly agree” to “strongly disagree”. For 2-way comparisons, we used Fisher’s exact test to compare the proportions of those five answer choices between different groups. Agree or strongly agree as well as disagree or strongly disagree were both collapsed into single categories. For multivariable comparisons, we used generalized linear models to assess how the demographic factors of gender (female vs male), cohort (current PSDP scholar vs PSDP alum) and URIM status (yes/no) correlated with the answers to the survey questions expressed as a numerical scale; we used binomial logistic models with the same predictors for categorical outcomes. Rather than asking individuals to self-identify, we asked specific demographic questions (Table I) and then applied the NIH definition of URIM.[14] Specifically, we classified individuals from Black/African American, Hispanic/Latino, or socioeconomically disadvantaged backgrounds and individuals with disabilities as URIM. We calculated all statistics using R version 4.0.5,[15] and we judged statistical significance using two-tailed tests at an alpha of 0.05.

Results

We sent email invitations to participate in the study to 55 individuals (21 current PSDP scholars and 34 alums), and we received a total of 126 responses to the survey: 41 in May 2020; 40 in August 2020; 24 in November 2020; and 21 in November 2021. Nineteen survey responses were from URIM individuals. Because the survey was anonymous, we could not identify the number of unique respondents or match multiple responses from the same individuals. Table II shows the demographics from all survey responses; the ratios of the demographic categories did not differ significantly between the overall numbers and between any of the individual time points.

Table 2.

Demographics from all survey responses.

Gender - All responses	n (%)
Female	54 (49.5%)
Male	54 (49.5%)
Other	1 (0.9%)
Gender - URIM only
Female	9 (47.4%)
Male	10 (52.6%)
Other	0 (0%)
Gender - non-URIM only
Female	41 (50.6%)
Male	40 (49.4%)
Other	0 (0%)
PSDP Cohort
Alumni	50 (47.2%)
Current scholars	56 (52.8%)
Race
Asian	13 (12%)
Black or African American	3 (2.8%)
White	80 (74%)
More than one race	9 (8.3%)
Prefer not to answer	3 (2.8%)
URIM
No	81 (81%)
Yes	19 (19%)

We found no significant differences (by Fisher’s exact test) between these percentages for all survey responses and between the percentages at the four different survey time points. Note that not all categories add up to the same total because some participants did not answer some questions.

Overall, 76.7% of survey responses either agreed or strongly agreed that they anticipated short-term delays in their research. That percentage was relatively stable throughout the first year of the pandemic — from May 2020 through November 2020, but it fell to 47.6% on the final round of surveys in November 2021 (Table III shows how the rate varied among survey timepoints.)

Table III.

Numbers and percentages of the different Likert scale responses at the four different survey time points.

		Survey
		5/2020	8/2020	11/2020	11/2021
Anticipating short-term delays (all respondents)	Agree or Strongly Agree	27 (81.8%)	26 (89.7%)	16 (80%)	10 (47.6%)
	Neither Agree nor Disagree	2 (6.1%)	0 (0%)	2 (10%)	5 (23.8%)
	Disagree or Strongly Disagree	4 (12.1%)	3 (10.3%)	2 (10%)	6 (28.6%)
Anticipating short-term delays (only URIM respondents)	Agree or Strongly Agree	5 (83%)	5 (100%)	3 (100%)	4 (80%)
	Neither Agree nor Disagree	1 (16.7%)	0 (0%)	0 (0%)	1 (20%)
	Disagree or Strongly Disagree	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Anticipating short-term delays (non-URIM respondents)	Agree or Strongly Agree	20 (80%)	19 (100%)	11 (73.3%)	6 (40%)
	Neither Agree nor Disagree	1 (4%)	0 (0%)	2 (13.3%)	3 (20%)
	Disagree or Strongly Disagree	4 (16%)	0 (0%)	2 (13.3%)	6 (40%)
Anticipating long-term delays (all respondents)	Agree or Strongly Agree	21 (64%)	19 (66%)	10 (50%)	11 (52%)
	Neither Agree nor Disagree	7 (21%)	5 (17%)	6 (30%)	4 (19%)
	Disagree or Strongly Disagree	5 (15%)	5 (17%)	4 (20%)	6 (29%)
Anticipating long-term delays (only URIM respondents)	Agree or Strongly Agree	4 (67%)	4 (80%)	3 (100%)	3 (60%)
	Neither Agree nor Disagree	2 (33%)	1 (20%)	0 (0%)	2 (40%)
	Disagree or Strongly Disagree	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Anticipating long-term delays (non-URIM respondents)	Agree or Strongly Agree	16 (64%)	13 (59%)	6 (40%)	7 (47%)
	Neither Agree nor Disagree	5 (20%)	4 (18%)	5 (33%)	2 (13%)
	Disagree or Strongly Disagree	4 (16%)	5 (23%)	4 (27%)	6 (40%)
Reconsidering their career (all respondents)	Yes	7 (21%)	8 (28%)	7 (32%)	7 (33%)
	No	26 (79%)	21 (72%)	15 (68%)	14 (67%)
Reconsidering their career (only URIM respondents)	Yes	4 (67%)	2 (40%)	1 (33%)	2 (40%)
	No	2 (33%)	3 (60%)	2 (67%)	3 (60%)
Reconsidering their career (non-URIM respondents)	Yes	3 (12%)	6 (27%)	5 (29%)	5 (33%)
	No	22 (88%)	16 (73%)	12 (71%)	10 (67%)
Changed their career path (all respondents)	Agree or Strongly Agree	3 (9%)	1 (3%)	1 (5%)	2 (10%)
	Neither Agree nor Disagree	1 (3%)	3 (10%)	4 (20%)	3 (14%)
	Disagree or Strongly Disagree	29 (88%)	25 (86%)	15 (75%)	16 (76%)
Changed their career path (only URIM respondents)	Agree or Strongly Agree	1 (17%)	1 (20%)	0 (0%)	1 (20%)
	Neither Agree nor Disagree	1 (17%)	1 (20%)	1 (33%)	1 (20%)
	Disagree or Strongly Disagree	4 (67%)	3 (60%)	2 (67%)	3 (60%)
Changed their career path (non-URIM respondents)	Agree or Strongly Agree	2 (8%)	0 (0%)	0 (0%)	1 (7%)
	Neither Agree nor Disagree	0 (0%)	2 (9%)	3 (20%)	2 (13%)
	Disagree or Strongly Disagree	23 (92%)	20 (91%)	12 (80%)	12 (80%)

Among survey responses from only URIM participants, 89.4% either agreed or strongly agreed that they anticipated short-term delays in their research, and that percentage was stable at all four surveys. On survey responses from non-URIM participants, however, only 72.7% anticipated short-term research delays, and that percentage trended down over the course of the study from 80% in May 2020 to only 40% in November 2021 (Table III). In multivariable linear regressions, we found that URIM status significantly correlated with whether the respondent anticipated short-term delays in their research (p <0.01); the respondent’s gender and PSDP cohort was not significantly associated with short-term research delays, and there were no significant interactions between any of the variables.

Overall, 59.2% of survey responses either agreed or strongly agreed that they anticipated long-term research delays due to the COVID-19 pandemic, and that rate was stable among the four surveys (Table III).

Among survey responses from only URIM participants, 73.7% either agreed or strongly agreed that they anticipated long-term delays in their research, and that percentage remained stable on all surveys. Among non-URIM survey respondents, only 54.6% anticipated long-term delays in their research, and that rate was also stable throughout the study (Table 3). The multivariable linear models showed that URIM status was significantly correlated with whether the respondent anticipated long-term delays in their research (p = 0.025); there was no significant relationship between respondent’s gender or PSDP cohort and long-term research delays, and there were no significant interactions between any of the variables.

Among all survey responses, 27.6% reported that they were reconsidering their career as a physician-scientist due to the impacts of COVID-19, and that percentage did not change much among the four survey timepoints (Table III). Separated by URIM status, 47.3% of responses from URIM participants reported reconsidering their career compared with only 24.0% of responses from non-URIM participants. Binomial logistic regression showed that both URIM background and male gender were significantly associated with respondents reconsidering a career as a physician-scientist (p = 0.026 and 0.044, respectively), and that URIM background had a stronger effect (β = 1.25 vs 1.005, respectively). PSDP cohort had no significant effect (p = 0.07), and there were no significant interactions between any variables.

Despite the high percentage of survey respondents who indicated they were reconsidering their career, only 6.8% of survey responses agreed or strongly agreed that they had actually changed their intended career path or specialty, and that percentage was stable throughout the study (Table III). Separated by URIM status, however, 15.8% of survey responses from URIM participants agreed or strongly agreed that they had changed their career path, compared with only 3.9% of responses from non-URIM participants. In the multivariable linear model, both URIM background and male gender were significantly associated with respondents changing their career path or specialty (p = 0.04 and 0.01, respectively); PSDP cohort had no significant effect, and there were no significant interactions between variables.

Discussion

The COVID-19 pandemic hampered research across many disciplines. In this study, we surveyed PDSP participants, and we focused particularly on individuals from URIM backgrounds. URIM physicians have long been doubly underrepresented among clinician-scientists, and in this small sample, we found that the adverse effects of the pandemic similarly disproportionately affected these trainees. Comparing surveys from URIM and non-URIM respondents, we found that URIM participants were significantly more likely to anticipate both short-term and long-term delays in their research, URIM participants were significantly more likely to be reconsidering a career as a physician-scientist, and URIM participants were significantly more likely to have changed their intended career path as a result of the pandemic. Those findings align with previous qualitative reports of disproportionate impacts from the pandemic on the careers of URIM physician-scientists [7,13] and highlight the need to support these individuals.[5] In addition, the disparity we observed is dramatically larger than those seen before COVID-19: a pre-pandemic review of MD-PhD students’ outcomes found that 26.9% of White students and 35.5% of URIM students dropped out of training [16]; in our data, 3.9% of non-URIM respondents reported leaving their career path compared with 15.8% of URIM respondents, despite being enrolled in a funded training program with robust mentorship and support.

In contrast to the effects we found on URIM respondents and the previous studies showing that women published less than men during the pandemic,[8–11,17] male respondents to our survey were significantly more likely than female respondents to report that they were reconsidering their career as a physician-scientist or had changed their intended career path. Our study differs from those previous reports in two notable ways. First, we focused our research in pediatrics, which may be a more family-friendly field than other medical specialties or scientific disciplines outside of medicine due to the high proportion of women in the field. Second, since we conducted this study within the context of PSDP, all the respondents on our surveys were already receiving intensive mentorship and support. Mentorship is a major factor in the early career success of URIM trainees,[4,18] so the benefits of PSDP may have tempered the challenges of the pandemic for the women participating in our study and blunted the disproportionate impact on women compared with men in this cohort.

One major challenge to interpreting and applying the results of our study is its size; we received only 126 responses to our surveys, and only 19 of them were from URIM individuals. Research with such a small sample risks both false-positive and false-negative results. Furthermore, we appreciate the challenges of interpreting research that runs the risk of overgeneralization. URIM physician scientists are a diverse group of individuals, and testing them as a singular group may miss the subtleties of their individual experiences. Unfortunately, the number of URIM individuals in the PSDP program is so small that considerations of research ethics compliance prevented us from identifying them any further; the unit of analysis would have been too small. Furthermore, because our survey was anonymous, we cannot report the number of unique individuals who responded or compare multiple responses from the same individual. We found statistically significant and concerning differences between URIM and non-URIM respondents, but we cannot ignore the possibility of sampling bias affecting our results. On the other hand, using anonymous surveys may have encouraged honest responses from participants, and such also facilitated rapid approval of this study; that allowed us to gather data while the effects of the pandemic were still ongoing and minimize recall bias.

Despite those challenges and the small size of this study, we believe this work highlights how the COVID-19 pandemic may have exacerbated the challenges faced by URIM physician-scientists. If we are to maintain a diverse and representative workforce in academic medicine and medical research, URIM trainees need ongoing and additional support from targeted development programs like PSDP.