Impact of COVID-19 Pandemic on Physician-Scientist Trainees to Faculty One Year into the Pandemic

Purpose Physician-scientists play a crucial role in advancing biomedical sciences. Proportionally fewer physicians are actively engaged in scientific pursuits, attributed to attrition in the training and retention pipeline. This national study evaluated the ongoing and longer-term impact of the COVID-19 pandemic on research productivity for physician-scientists at all levels of training. Methods A survey of medical students, graduate students, and residents/fellows/junior faculty (RFJF) was conducted from April to August 2021 to assess the impact of COVID-19 on individual stress, productivity, and optimism. Multivariate regression analyses were performed to identify associated variables and unsupervised variable clustering techniques were employed to identify highly correlated responses. Results A total 677 respondents completed the survey, representing different stages of physician-scientist training. Respondents report high levels of stress (medical students: 85%, graduate students: 63%, RFJF: 85%) attributed to impaired productivity concerns, concern about health of family and friends, impact on personal health and impairment in training or career development. Many cited impaired productivity (medical students: 65% graduate students: 79%, RFJF: 78%) associated with pandemic impacts on training, labs closures and loss of facility/resource access, and social isolation. Optimism levels were low (medical students: 37%, graduate students: 38% and RFJF: 39%) with females less likely to be optimistic and more likely to report concerns of long-term effects of COVID-19. Optimism about the future was correlated with not worrying about the long-term effects of COVID-19. Since the COVID-19 pandemic, all respondents reported increased prioritization of time with family/friends (67%) and personal health (62%) over career (25%) and research (24%). Conclusions This national survey highlights the significant and protracted impact of the COVID-19 pandemic on stress levels, productivity, and optimism among physician-scientists and trainees. These findings underscore the urgent need for tailored support, including mental health, academic, and career development assistance for this biomedical workforce.


Introduction
Despite the unique role of physician-scientists in advancing discovery and application of biomedical science to human health, several previous reports suggest that the current number of physician-scientist trainees may not be su cient to meet expected workforce needs. 1,2.The physician-scientist training pipeline is long and includes many unique challenges, including nancial stressors caused by increasingly delayed matriculation into the workforce, pressure to balance both clinical and scienti c productivity, and persistent competition for research funding.In addition to these existing stressors, the presence of additional factors related to COVID-19 may negatively add to the strain experienced by trainees and early career physician-scientists.Given the multiple critical junctures in the physicianscientist training pipeline, it is important to study the effect of these disruptive events and any potential impact they may have contribute on attrition in this already vulnerable group.
We previously reported the results of a national survey examining the effect of COVID-19 on the personal life, career, stress, and productivity amongst physician-scientists at different stages of the training and early-career pipeline (medical student, graduate student, resident, fellow, and early faculty). 3Our prior work found that all respondent groups reported high levels of stress, social isolation, and negative impact on productivity resulting from the pandemic.These adverse experiences were disproportionately more likely to be described by women, individuals from disadvantaged backgrounds, and groups traditionally underrepresented in medicine (URM) including Black or African Americans, Hispanic or Latinos, American Indians or Alaskan Natives, Native Hawaiians, and other Paci c Islanders.In addition, some negative consequences of the pandemic, such as social isolation and nancial di culties, were found to be particularly prevalent and disruptive.
Our prior survey data were collected between April and June 2020 3 , capturing immediate and early consequences of the COVID-19 pandemic on trainees and early-career physician-scientists.While our prior work was important for characterizing the early impact of the pandemic, the longer-term consequences of the pandemic and subsequent national response are not known.Understanding the persistent impact of the pandemic on physician-scientist trainees is crucial in order to better support this workforce and potentially mitigate further attrition in this already "leaky" pipeline.Here, we present the results of a national follow-up survey that evaluates the ongoing and longer-term consequences of the COVID-19 pandemic on the personal and professional lives of physician-scientists at all levels of training.

Survey Design and Recruitment
The survey tool was designed with feedback from mental health researchers and academic faculty with expertise in training physician-scientists. Three versions of the survey were tailored to physician-scientist training level for medical students, graduate students, and residents/fellows and junior faculty (RFJF).This study was reviewed and approved as exempt by the Weill Cornell Medicine Institutional Review Board.
From April 2021 to August 2021, the survey was distributed to 120 U.S. institutions with MD-PhD programs by the chairs of the Association of American Medical Colleges' Group on Graduate Research, Education and Training and by institutional representatives of the American Physician Scientists Association.The survey was administered via SurveyMonkey (Momentive, San Mateo, California).

De ning Response Variables
For each group of respondents (medical students, graduate students, and RFJF), three outcome measures were de ned as an aggregate of response to particular survey questions.Stress outcome was de ned as positive if a respondent indicated "Agree" or "Strongly Agree" to any of the following statements: 1. "The COVID-19 pandemic has caused me sleep problems, decreased energy, changes in appetite, di culty concentrating, and/or restlessness" 2. "The COVID-19 pandemic has caused me a signi cant amount of stress, anxiety, hopelessness, and/or depression" 3. "Uncertainty of not being able to nish my research or to graduate is a great source of stress" Productivity outcome was de ned as positive if a respondent indicated "Agree" or "Strongly Agree" to any of the following statements: 1. "My research productivity/medical training will be negatively impacted in the long term" 2. "My research productivity/medical training will be negatively impacted in the short term" Optimism outcome was de ned as positive if a respondent indicated "Agree" or "Strongly Agree" to the following statement: 1. "I am optimistic about the future given the trajectory of the COVID-19 pandemic"

Multivariate Regression Analyses
Each outcome metric in every respondent group was evaluated individually.All variables were reduced to those associated with outcome by Elastic-Net Regularized Regression using the glmnet package in R v3.6.1.This feature set was further reduced by backwards stepwise feature selection, maximizing the Akaike Information Criterion (AIC).Statistical signi cance of each variable in the reduced feature set and odds ratio with respect to outcome were then assessed by multivariate Generalized Logistic Regression Model.Variables were also ranked by importance in a random forest model according to mean decrease in accuracy for each variable.Overall model predictive accuracy was quanti ed by Area-Under-the-ROC-Curve (AUC).

Unsupervised Variable Clustering
For each group of survey respondents, supervised multivariate analyses of variables in association with each outcome were performed.Additionally, an unsupervised clustering of all survey response variables was completed in order to assess for "blocks" of highly correlated survey responses.Distance between variables was assessed by Spearman correlation, and clustering was performed by Partitioning Around Medioids (PAM), with optimal number of clusters selected by maximization of mean silhouette score.Pairwise statistical signi cance of association between variables was also assessed by Fisher's Exact test.Separate heatmaps of variable-to-variable correlation were generated for each group of survey respondents across 1) all variables and 2) only the set of variables with pairwise p-values < 0.05, following Benjamini-Hochberg multiple-testing correction.This enabled direct visualization of highly correlated sets of survey responses in unbiased fashion.
In a multivariate model for the outcome of stress, those who were working from home (OR 2.42, 1.15-5.22,p = 0.021), those who indicated their productivity will be negatively affected in the long term (OR 2.63, 1.07-7.14,p = 0.043) and those worrying about the health of family and friends (OR 2.97, 1.43-6.34,p = 0.004) were more likely to report increased stressed during the COVID-19 pandemic (Fig. 1a).The perception that virtual encounters are as good as in-person encounters for medical training was protective against stress among MS respondents (OR 0.18, 0.03-0.82,p = 0.035) (Fig. 1a).
Regarding productivity, a total of 117 (65%) MS respondents described their productivity as being adversely impacted by the pandemic ( (Fig. 1b).
In terms of optimism, female respondents were signi cantly less likely to report being optimistic than their male counterparts (n = 25, 37.9% vs n = 32, 64%, p = 0.005) (Table 3).Other than gender, the optimistic and non-optimistic cohorts did not differ by any other demographic or training characteristics.Respondents who were less optimistic were more likely to attribute feelings of stress, anxiety, and hopelessness due to the pandemic (n = 27, 40.9% vs n = 30, 60%, p = 0.042), describe depressive symptoms such as problems with sleep, fatigue, and changes in appetite (n = 25, 37.9% vs n = 30, 60%, p = 0.018), and report concern for the health of their friends and family (n = 34, 51.5% vs n = 37, 74%, p = 0.014) (Table 3).Of note, respondents who were less optimistic about the pandemic also described feeling stress due to many of the policy and political efforts in response to the pandemic, including concern regarding health disparities and gender inequity (n = 38, 57.6% vs n = 40, 80%, p = 0.011), and how the pandemic was managed at the local and national levels.Finally, there was greater concern regarding the potential long-term impact of COVID-19 on the careers and personal lives of those who were less optimistic about the pandemic (n = 22, 33.3% vs n = 33, 66%, p = < 0.001) See table 2 for characteristics of those who were more optimistic (  1).
The COVID-19 pandemic produced several academic and social stressors for this population.More than half of the respondents noted that they worry about the long-term effects that COVID-19 will have on their career, personal life, and/or family/friends (n = 172, 53.9%).One-third of respondents (n = 106, 33.2%) had their lab shut down, while 80.6% (n = 257) had experiments delayed or impaired and 11.6% (n = 37) were unable to perform any research activities.Important milestones were postponed: 41 (12.9%) respondents had their qualifying or thesis exam postponed and 45 (14.1%) had their transition back to medical school delayed.74% of respondents (n = 236) agreed that their research productivity and/or medical training was being negatively impacted in the short-term, and 49.8% (n = 159) believed that it would affect their career in the long-term.Almost two-thirds of the respondents experienced a signi cant amount of stress, anxiety, hopelessness and/or depression (n = 201, 63%).This included 191 (54.5%) who reported sleep problems, decreased energy, changes in appetite, di culty concentrating and/or restlessness (Supplemental Table 1).
In a multivariate regression analysis, those who indicated spending time enhancing career development through online resources were less likely to be stressed (OR 0.38, 0.18-0.82,p = 0.012).Those who were dual degree students (OR 3.69, 1.09-12.47,p = 0.033) or those who were physically isolated due to work (OR 4.45, 2.39-8.62,p = < 0.001) were more likely to be stressed (Fig. 2a).
In addition to affecting trainee stress levels, the COVID-19 pandemic had a signi cant and negative impact on GS training, productivity, and career development.The vast majority of students felt that virtual classrooms failed to adequately recapitulate in-person learning (83.5%, n = 248).While the majority of students reported that their labs did not shut down (65.0%, n = 193), most students did report that experiments were delayed due to the COVID-19 pandemic (82.2%, n = 244).Nearly all GS respondents felt that their research productivity would be impacted in the short term (79.5%, n = 236), and a signi cant proportion also felt that this impact would extend into the long-term (53.5%, n = 159) (Supplemental Table 1).
Interestingly, the COVID-19 pandemic had a signi cant effect on GS research topic of choice or intended career path for a subset of respondents.Thirty-one (9.7%) respondents shifted their research efforts to COVID-19 related topics, while twenty-six (8.2%) shifted their intended career path due to the pandemic.Expectedly, the overwhelming majority of students whose productivity was adversely effected reported as being stressed by the pandemic (90.4%, n = 224).Consistent with the stress and encumbered productivity among GS respondents, only 121 (55%) of students reported feeling optimistic (Table 3).In a multivariate analysis, being at a public institution (OR 1.98, 1.21-3.27,p = 0.007) and spending time performing computational modeling (OR 2.14, 1.15-4.0,p = 0.016) were associated with optimism for the future (Fig. 2c).

Resident, Fellow, Junior Faculty (RFJF)
We received 178 responses from physician-scientists in post-graduate medical training and early career faculty positions, with an average age of 35.7 ± 5.7 years (range 19-51), 51% women, 9% Hispanic/Latino(a), and 3% Black/African American.Of note, the proportion of respondents for the latter two groups is similar to their proportion among MD-PhD graduates during this period. 4Among respondents, 74% were married or had a committed partner, and 48% had children.Completion of a formal dual degree training was reported in 60% of respondents, which is a signi cantly smaller proportion than represented amongst our medical and graduate student respondents (p < 0.001), while 23% were in a tenure track position (Table 1 and Supplemental Table 1).
Factors most predictive of stress during the pandemic, and which were reported by 85% of subjects, included research laboratory shutting down (OR 5.  3, Fig. 3c).Finally, we found that respondents at every level of physician-scientist training and early career reported prioritization of friends/family (66%), personal health 62%, time with partner (48%) and exercise (40%), whereas career success was reported by only 25% and research by 24% of respondents (Fig. 4).

Discussion
This national survey of physician-scientists and trainees provides key insights into the personal, psychosocial, educational, and professional implications of the COVID-19 pandemic in the United States.This work is a follow-up to a prior national survey conducted by our group and is meant to provide a cross-sectional point of reference of the major COVID-19 related stressors impacting the training and careers of physician-scientists. 3 The data presented here reveal that the pandemic continues to have a signi cant effect on the stress, productivity, and perceived optimism of trainee and early career physicianscientists.Despite extensive prior work detailing these impacts, this survey suggests that many of the initial adverse consequences of the pandemic continue to have a persistent effect in this cohort.Taken together, the ndings of the present study suggest that there is an ongoing need to address such concerns in a manner that is actionable and tailored to the speci c training stage of early career physician-scientists and trainees.

Medical students (MS)
Initial reports following the onset of COVID-19 described early adverse effects on MS respondents, including increased sense of social isolation, stress, and limited academic and scholarly productivity. 3,5,6hile the negative impact of the COVID-19 pandemic was found to be consistent across all MS demographic cohorts, prior studies suggest that the effects were most severe amongst trainees who identi ed as female and URM. 3,6These trends raised concern that the differential impact of the pandemic amongst trainees may further exacerbate career disparities for groups traditionally underrepresented in medicine.9][10] MS responses to this survey underscore the continued negative impact of COVID-19 on stress, productivity, and optimism.Multivariate regression modeling demonstrated that working from home, impaired productivity, and concerns regarding the health of friends/family were each independently associated with increased feelings of stress.Importantly, the only protective factor was the perception that virtual clinical encounters were adequate for supporting medical training.These responses suggest that while the factors associated with greater feelings of stress among medical students are diverse and span both personal and professional domains, limitations of hybrid or virtual education formats may provide an actionable area of improvement for mitigating stress in this cohort.][9][10] The results of this survey suggest that despite the necessity of virtual clinical curricula early in the pandemic, perceived limitations of virtual clinical education have a negative impact on the student learning experience.Reinstitution of in-person clinical and academic learning is likely to be bene cial for improving clinical con dence among trainees.
In addition to the direct personal and professional stressor caused by the pandemic, the indirect impact of the pandemic on productivity appears to be persistent among MS trainees.Unsurprisingly, those who described their clinical and academic progression as being signi cantly delayed reported greater levels of impaired productivity.While this survey did not evaluate the timing and extent of academic delays in this cohort, the nding that such trainees are experiencing delays is particularly relevant given the already increasing time to graduation among physician-scientist trainees.Interestingly, students attending public institutions were less likely to describe impaired productivity.It is not clear whether the protective effect of public-school attendance is due to selection bias among respondents or a re ection of the quality of the institutional responses to the pandemic, including better educational, academic, or psychosocial support resources.This is certainly an area that warrants additional investigation.
Finally, looking to the future of the pandemic, our results demonstrate a heterogenous response in perceived optimism among MS respondents.Optimistic respondents were more likely to be male, those who were able to engage in COVID-19 related research, and those who did not believe the timeline of their academic progression to be negatively impacted.The more optimistic perception of the future of the pandemic among males is consistent with our prior survey showing that the elevated stress early in the pandemic tended to be disproportionately impacting female MS. 3 Prior studies have shown that the rates of self-reported depression were greater among female MS following the initial onset of the pandemic. 11t is likely that the reduced optimism described here is a sequela of this persistent depressive effect of the pandemic.This raises concern that the ongoing response to the pandemic has not been su cient to support female trainees and demands greater understanding of the unique obstacles experienced by this cohort.Moreover, our survey results suggest that beyond female respondents, there continues to be a larger contingent of medical students who expressly state concern about the lasting impact of COVID-19 on their academic and professional careers, which warrants additional attention by academic and professional leadership.

Graduate students (GS)
Prior studies on the GS experience during COVID-19 highlighted signi cant effects of the pandemic on mental health and academic productivity.Reported stressors early in the pandemic included isolation from friends and family, uncertainty about the future, and di culty with social distancing. 113][14][15] Biomedical GS researchers in particular cited time management and decreased productivity as stressors during the COVID-19 pandemic, likely owing to the unique di culties of conducting wet lab research during periods of mandated restriction to in-person activities. 12,16Multiple student commentaries during this time cite the obstacles in adapting new methods of maintaining research productivity in order to promote future career advancement. 17,18e results of the present survey clearly reveal that the COVID-19 pandemic continues to meaningfully affect GS trainees.Most GS respondents feel that their productivity-and in turn, their careers-have been delayed or impacted in both the near-and long-term.Despite a greater concern on productivity among students involved in non-computational research, negative perceptions regarding long-term productivity were shared across both wet and dry lab researchers.This is an impact that will necessarily warrant continued assessment for many years to come.Such gaps in trainee productivity and decreased optimism about the future will likely be negatively re ected in subsequent applications for training grants, career development awards, primary research funding, and post-doctoral or faculty applications.
Additionally, this survey con rms the pandemic intensi ed the stress felt by graduate trainees.Namely, our study emphasizes that the amount of stress the COVID-19 pandemic added to trainees' lives is paramount to consider with careful intention.As medical and graduate trainees' mental health and wellbeing are already known to be at risk-and often overlooked-the added stress of the pandemic has only exacerbated these issues. 19,20Indeed, over half of respondents indicated they experience physical symptoms of increased stress, depression, or anxiety greater than one year from the initial onset of the pandemic.With already established trends in worsening mental health among health care trainees and professionals, these ndings indicate that continued attention must be paid to ensure that the well-being of physician-scientist trainees is supported.
Resident, Fellow, Junior Faculty (RFJF) In addition to high levels of stress and lost productivity, RFJF were more likely than MS and GS respondents to report nancial di culties early in the pandemic. 3Those with a dual degree reported more impairment in productivity than their non-dual degree holding counterparts. 3This appeared to be due, at least in part, to increased clinical duties resulting from increased demands on the healthcare system, as well as supply chain disruptions leading to limited access to essential research resources.In addition, women were more likely than their male counterparts to have increased home demands due to taking care of children, leading to further decreased productivity.
In the present survey, RFJF continued to report high levels of stress and loss of productivity because of the COVID-19 pandemic.These factors could potentially have devastating consequences for a generation of physician-scientists at a crucial point of their career.Indeed, many physicians abandon their research career at the stage of early career awards, notably at the transition from K to R awards. 21Stress and loss of productivity can also be compounded by the aforementioned worsening nancial pressure.
Interestingly, hematology/oncology respondents from this group were less optimistic about the future.While we did not identify a de nite reason for this association, multiple factors including the need for frequent in-person treatment visits, lapses in cancer treatment and screening caused by the pandemic, and the immunocompromised nature of their patient population may be contributory. 22However, these ndings may also be due to a selection bias in the subset of respondents sampled from this specialty.Irrespective of specialty-speci c experiences, the ndings from this survey indicate that personal and professional stressors related to the pandemic continue to have a protracted impact on RFJF.Such factors certainly add to the challenges experienced by this cohort as they attempt to establish their careers as independent investigators, and the consequences of these additive challenges need to be further studied.

Figure 1 Medical
Figure 1

Figure 2 Graduate
Figure 2