Abstract
BACKGROUND
Resident physicians provide front-line care to coronavirus disease 2019 (COVID-19) patients, but little is known about how they perceive the risk to their own health or how this is affected by the increasing role of social media in disseminating information. This study aims to determine resident physicians’ perceptions of personal COVID-19 risk during the first COVID wave and compare risk perceptions between low–average and high social media users.
METHODS
We conducted a cross-sectional survey at the University of Toronto in May 2020 among resident physicians in internal medicine, emergency medicine, critical care, and anaesthesia. Participants were considered high social media users if above the median for daily social media use and low–average users if at or below the median. The primary outcome was perceived risk of hospitalization with COVID-19 within 6 months.
RESULTS
A total of 98 resident physicians reported a median of 1–2 hours daily on social media, and 55.7% endorsed social media as a very or the most common source of information on COVID-19. The median overall perceived risk of hospitalization was 10% (inter-quartile range [IQR] 5–25)—7.5% for low–average social media users and 17.5% for high social media users (p = 0.10).
CONCLUSIONS
Resident physicians have an elevated perception of COVID-19 risk, including a perceived risk of hospitalization 250 times greater than the local population risk. Although social media are an important source of information on COVID-19, risk perception did not significantly differ between high and low–average social media users.
Keywords: COVID-19, medical education, risk perception
Abstract
HISTORIQUE
Les résidents en médecine donnent des soins de première ligne aux patients atteints de la maladie à coronavirus 2019 (COVID-19), mais on possède peu d’information au sujet de la perception de leur risque personnel ou de l’effet de la diffusion croissante d’information et de désinformation dans les réseaux sociaux sur leur perception. La présente étude vise à déterminer les perceptions du risque personnel de COVID-19 chez les résidents en médecine pendant la première vague de la pandémie et à comparer les perceptions de risque entre les utilisateurs faibles à modérés et les grands utilisateurs des réseaux sociaux.
MÉTHODOLOGIE
En mai 2020, les chercheurs ont réalisé une étude transversale auprès des résidents en médecine interne, en médecine d’urgence, en soins intensifs et en anesthésie de l’Université de Toronto. Ceux qui se situaient au-dessus de la médiane d’utilisation quotidienne des réseaux sociaux étaient considérés comme de grands utilisateurs des réseaux sociaux, et ceux qui se situaient sur ou sous la médiane, comme des utilisateurs faibles ou modérés. Le résultat clinique primaire était le risque perçu d’hospitalisation à cause de la COVID-19 dans les six mois, et les résultats cliniques secondaires, le risque estimatif de contracter la COVID-19 ou d’infecter des membres de la famille, le degré d’anxiété au sujet de la COVID-19 et les répercussions perçues de l’hospitalisation.
RÉSULTATS
Au total, 98 résidents en médecine ont déclaré consacrer une médiane de une à deux heures par jour aux réseaux sociaux, et 55,7 % considéraient les réseaux sociaux comme leur principale source d’information ou une source d’information très importante sur la COVID-19. Le risque médian perçu de contracter la COVID-19 s’élevait à 60 % (ratio interquartile [RIQ] 32 à 75), d’être hospitalisé, à 10 % (RIQ 5 à 25), et d’infecter des membres de la famille, à 37 % (RIQ 10,5 à 60). Le risque médian perçu d’hospitalisation s’élevait à 7,5 % pour les utilisateurs faibles ou modérés des réseaux sociaux, et à 17,5 % pour les grands utilisateurs des réseaux sociaux (p = 0,10).
CONCLUSIONS
Les résidents en médecine ont une perception élevée du risque de COVID-19, y compris une perception du risque d’hospitalisation 250 fois plus élevée que celle de la population locale. Même si les médias sociaux représentent une source importante d’information sur la COVID-19 pour les résidents en médecine, la perception du risque ne différait pas de manière significative entre les grands utilisateurs et les faibles utilisateurs des réseaux sociaux.
Mots-clés : COVID-19, formation médicale, perception du risque
Introduction
On February 15, 2020, the general director of the World Health Organization commented, “We’re not just fighting an epidemic, we are fighting an infodemic” (1,2). The increasing popularity and importance of health information online and on social media is outpacing traditional methods of health knowledge translation (1,3).
In early January 2020, worldwide news sources began reporting on a rapidly spreading new coronavirus in China, triggering an avalanche of information searches on the Internet (1). The limited availability of information created a void rapidly filled by poor-quality information. Some social media platforms have attempted to limit false information or redirect users to reliable resources, but they remain a conduit for misinformation (2). Users may not critically assess the information they read, and online misinformation can influence perceptions of risk and affect behaviours (4,5).
A common perception is that educated health care workers are impervious to online health misinformation. However, social media have taken on increasing importance in the dissemination of information and research in health care, as evidenced by the rising popularity of Free Open Access Medical Education networks, “tweetorials,” and the sharing of pre-print academic publications on social media (6). Although useful for rapid knowledge translation, there are potential risks to these non–peer-reviewed materials (6–8). At the start of the pandemic, attempts to rapidly translate the most up-to-date information to front-line workers outpaced traditional scholarly publications and clinical tools, and social media increasingly filled the gap (6,9). Although physicians are typically assumed to have the knowledge to critically assess information, much of the information on coronavirus disease 2019 (COVID-19) is subspecialized and outside the training of many physicians affected by the pandemic.
Resident physicians are exposed to social media in both medical and general public social media circles. We sought to describe residents’ perceptions of their personal COVID-19 risk during the initial wave and to compare perceptions of risk between high and low–average social media users. We hypothesized that resident physicians would have elevated and widely varying personal estimates of COVID-19 risk and that these estimates would be greater among high social media users.
Methods
General study design
We conducted a cross-sectional survey with the aim of describing resident physicians’ perceptions of their personal COVID-19 risk and also tested for an association between resident risk perception and social media use.
Survey design
The survey was designed in accordance with recommendations from the Academy of Critical Care: Development, Evaluation and Methodology Group (10). Risk perception was measured using a tripartite structure with probabilistic, affective, and consequential dimensions that explains variance in health behaviours better than probabilistic dimensions alone (11). The survey was subjected to pilot and sensibility testing with six University of Toronto chief medical residents.
Study administration and participants
From April 30 to May 12, 2020, we invited University of Toronto residents in anaesthesia, critical care, emergency medicine, and internal medicine to participate in a self-administered online questionnaire. These specialities were chosen for their direct involvement in the care of COVID-19 patients. A recruitment e-mail was sent to residents enrolled in each program, detailing the voluntary nature of participation and providing a brief background, a declaration of anonymity, and the link to the questionnaire.
Data collection
The survey consisted of three parts: basic demographic characteristics, perceptions of risk, and social media and other resource use. Demographic data included age, sex, program, postgraduate year (PGY), and household members. The risk perception questions were asked before the social media and resource use questions to minimize framing bias.
Primary exposure: Social media use
To assess social media and other information sources, participants were asked to quantify time spent daily on social media and on research and online clinical tools both currently and in the 12 months preceding the pandemic. In addition, they were asked to rank the usefulness of each source of information on a Likert scale ranging from 1 (not a source of information used) to 5 (most common source of information).
After the survey data were collected, the distribution of current social media use was inspected, and on the basis of the median, respondents were dichotomized into high (above the median) and low–average (at or below the median) social media use. This process was repeated for primary research use and online clinical tools.
Primary outcome: Perceived risk of hospitalization
Participants were asked to provide an estimate of their probability of (a) acquiring COVID-19, (b) being hospitalized with COVID-19, and (c) infecting family members with COVID-19, each within the next 6 months. Participants were also asked to rate their degree of anxiety on a Likert scale ranging from 1 (no anxiety at all) to 5 (constantly very anxious) and the expected severity of impact that hospitalization would have on their lives on a Likert scale ranging from 1 (minimal lasting impact) to 5 (long-lasting impact) (11).
We selected the perceived risk of hospitalization as the primary outcome because there is more objective information available on the observed incidence and risk of hospitalization, whereas the general incidence of COVID-19 infection has been debated, particularly given high rates of undetected, asymptomatic carriage.
Statistical analysis
Descriptive statistics (medians with quartiles and frequencies) were calculated for all demographic characteristics and risk perception responses. Associations between each information source and the risk perception outcomes were examined using the Mann–Whitney Wilcoxon test. The statistical significance level was set at p < 0.05 (two-sided).
The primary analysis compared the perceived risk of hospitalization among those with low–average versus high social media use. In the secondary analyses, we compared other risk outcomes between these two groups. A sensitivity analysis was conducted by redefining low social media use as the first quintile and high social media use as the fifth quintile.
A pre-specified sub-group analysis was then conducted among those who ranked social media as their most common or very common source of information (4 or 5 on the Likert scale) and those who did not (1–3 on the Likert scale).
Results
Survey respondents
Of the 487 residents invited to participate, 98 responded (20.1% response rate), and 8 were excluded as incomplete, yielding a total of 90 survey respondents (Table 1). Participants’ median age was 28.5 (inter-quartile range [IQR] 27–31), 50 (55.6%) were female, and 64 (71.1%) shared a household with others. The majority of the respondents (n = 66; 73.3%) were from internal medicine, which reflects the larger size of this training program. The respondents were nearly evenly distributed among the four PGY categories (years 1, 2, 3, and ≥4).
Table 1:
Characteristics of respondents
| Characteristic | No. (%)* | p-value | ||
|---|---|---|---|---|
| Overall; N = 90 | Social media use | |||
| Low to average; n = 60 | High; n = 30 | |||
| Age, median (IQR) | 28.5 (27–31) | 29 (27–32) | 28 (26–30) | 0.12 |
| Sex | 0.55 | |||
| Male | 40 (44.4) | 28 (46.7) | 12 (40.0) | |
| Female | 50 (55.6) | 32 (53.3) | 18 (60.0) | |
| Training program | 0.02 | |||
| Anesthesia | 13 (14.4) | 11 (18.3) | 2 (6.7) | |
| Critical care | 6 (6.7) | 6 (10.0) | 0 (0) | |
| Emergency medicine | 5 (5.6) | 5 (8.3) | 0 (0) | |
| Internal medicine | 66 (73.3) | 38 (63.3) | 28 (93.3) | |
| PGY level | 0.43 | |||
| 1 | 23 (25.6) | 13 (21.7) | 10 (33.3) | |
| 2 | 20 (22.2) | 12 (20.0) | 8 (26.7) | |
| 3 | 23 (25.6) | 17 (28.3) | 6 (20.0) | |
| ≥4 | 24 (26.7) | 18 (30.0) | 6 (20.0) | |
| Household contacts | 0.51 | |||
| Any | 64 (71.1) | 44 (73.3) | 20 (66.7) | |
| Spouse or partner | 48 (53.3) | 34 (56.7) | 14 (46.7) | |
| Parents | 7 (7.8) | 5 (8.3) | 2 (6.7) | |
| Children | 7 (7.8) | 5 (8.3) | 2 (6.7) | |
| Siblings | 7 (7.8) | 3 (5.0) | 4 (13.3) | |
| Nonrelative | 7 (7.8) | 4 (6.7) | 3 (10.0) | |
| Grandparents | 1 (1.1) | 0 (0) | 1 (3.3) | |
| Other family member | 0 (0) | 0 (0) | 0 (0) | |
| Prior media use, median | ||||
| Social media | 1–2 h | 45–59 min | 2–3 h | |
| Research | 15–29 min | 15–29 min | 15–29 min | 0.84 |
| Clinical tools | 30–44 min | 30–44 min | 30–44 min | 0.68 |
| Current media use, median | ||||
| Social media | 1–2 h | 1–2 h | 2–3 h | |
| Research | 15–29 min | 15–29 min | 15–29 min | 0.42 |
| Clinical tools | 30–44 min | 15–29 min | 30–44 min | 0.15 |
*Unless otherwise indicated
IQR = Inter-quartile range; PGY = Postgraduate year
Social media use
The median current daily social media use was 1–2 hours (Figure 1) compared with 15–29 min for current primary research use and 30–44 min for online clinical tools. Using the median (1–2 hours) as the cut-point, 60 (66.7%) residents were considered low to average social media users, and 30 (33.3%) were considered high social media users. Demographic characteristics and use of primary research and online clinical tools were similar between the high and low–average social media groups, with the exception of training program because high social media users were more likely to be in internal medicine than in other specialities (Table 1). Of the 90 survey respondents, 88 completed the survey question about the usefulness of social media as a source of information, and 49 (55.7%) rated social media as the most common or very common source of information.
Figure 1:
Current daily social media use
Perception of risk
The overall median perceived risk of acquiring COVID-19 in the next 6 months was 60% (IQR 32–75), with a range from 10% to 100% (Figure 2A). The overall median perceived risk of hospitalization with COVID-19 in the next 6 months was 10% (IQR 5–25), with a range from 0% to 80% (Figure 2B). The overall median perceived risk of infecting family members with COVID-19 was 37% (IQR 10.5–60) with a range of 0% to 100% (Figure 2C).
Figure 2:
Risk perception outcomes stratified by social media use: (A) perceived risk of acquiring COVID-19 in the next 6 months, (B) perceived risk of hospitalization with COVID-19 in the next 6 months, and (C) perceived risk of infecting family members in the next 6 months
The overall median Likert-scale rating for anxiety about contracting COVID-19 was 3 (IQR 2–3). The overall median Likert-scale rating for perceived severity of impact on life if hospitalized with COVID-19 was 3 (IQR 3–4).
Association of social media use and COVID-19 risk perception
The median perceived risk of hospitalization was 7.5% (IQR 5–20) among low–average social media users and 17.5% (IQR 5–30) among high social media users (p = 0.10) (Table 2). In a sensitivity analysis comparing the lowest and highest quintiles of social media users, the perceived risks of hospitalization were 7.5% (IQR 5–20) and 17.5% (IQR 5–30), respectively (p = 0.235).
Table 2:
Perception of risk by social media use
| Outcome | Median (IQR) | p-value | ||
|---|---|---|---|---|
| Overall; N = 90 | Social media use | |||
| Low to average; n = 60 | High; n = 30 | |||
| Primary outcome: perceived risk of hospitalization with COVID-19 | 10 (5–25) | 7.5 (5–20) | 17.5 (5–30) | 0.10 |
| Secondary outcomes | ||||
| Perceived risk of acquiring COVID-19 | 60 (32.35–75) | 60 (30–75) | 60 (47.5–76.25) | 0.56 |
| Perceived risk of infecting family members | 37 (10.5–60) | 30 (10–60) | 50 (17.5–70) | 0.20 |
| Anxiety about COVID-19* | 3 (2–4) | 3 (2–3) | 3 (2–4) | 0.17 |
| Severity of impact on life of hospitalization* | 3 (3–4) | 3 (3–4) | 4 (3–4) | 0.43 |
*Rated on a scale ranging from 1 to 5
IQR = Inter-quartile range; COVID-19 = Coronavirus disease 2019
The median perceived risk of acquiring COVID-19 was 60% (IQR 30–70) among low–average social media users and was also 60% (IQR 47.5–76.25) among high social media users (p = 0.56). The median perceived risk of infecting family members was 30% (IQR 10–60) among low–average social media users and 50% (IQR 17.5–70) among high social media users (p = 0.20). The median Likert scale rating for anxiety about COVID-19 was 3 (IQR 2–3) among low–average social media users and 3 (IQR 2–4) among high social media users (p = 0.17). The median Likert scale rating for severity of impact of hospitalization was 3 (IQR 3–4) among low–average social media users and 4 (IQR 3–4) among high social media users (p = 0.43).
Perceived usefulness of social media
In the subgroup that rated social media as their most common or very common source of information, the median perceived risk of acquiring COVID-19 was 60% (IQR 30–80), the median perceived risk of hospitalization was 10% (IQR 5–30), and the median perceived risk of infecting family members was 40% (IQR 10–60). The median Likert scale rating for anxiety was 3 (IQR 2–4); for impact on life, it was 4 (3–4). In this subgroup, there was no statistically significant difference in any of the risk perception measures between low–average and high social media users.
Discussion
In this study of resident physicians involved in front-line care of COVID-19 patients, the median perceived risk of acquiring COVID-19 in the next 6 months was 60%, the median perceived risk of hospitalization with COVID-19 was 10%, and the median perceived risk of infecting family members was 37%. These numbers are drastically higher than those documented in the literature. On April 30, 2020 (the time of the study), and October 31, 2020 (6 mo after study), the cumulative incidences were, respectively, 0.11% and 0.52% in Ontario and, specifically, 0.16% and 0.87% in Toronto. The cumulative incidence of hospitalization in Ontario was 0.013% on April 30, 2020, and 0.04% on October 31 (12,13). Residents predicted a more than 750 times greater risk of hospitalization than the local population risk at the time and a 250 times greater risk than the eventual 6-month population risk. Despite several studies on COVID-19 among health care workers, the data on incidence are limited. By April 30, Ontario reported 2,419 cases among health care workers, with an estimated 870,000 health care workers in Ontario at the time, but it stopped reporting total health care worker infections by June 2020 (12,14). A survey of residency programs in New York reported an 11.6% infectivity rate among resident physicians in the first wave, but not all were confirmed cases (15). A meta-analysis of 46 studies conducted before July 8, 2020, found an estimated pooled prevalence of 11% among 75,859 health care workers; however, the individual studies varied widely by region (16). The variance is expected, depending on the baseline incidence of COVID-19 in the area and compounded by additional factors such as lack of personal protective equipment, hospital overcrowding, and local health care systems being overwhelming. Yet the infectivity rate among health care workers was still several orders of magnitude less than the risk perceived by resident physicians in our survey.
There are many possible reasons why residents would have inflated perceptions of risk. Although case numbers are frequently reported in the media, information on prevalence and incidence are rarely reported. The media also report that health care workers are at elevated risk, with emotionally charged anecdotes of young health care workers being infected (17). Residents are therefore inundated with rising case numbers and warnings that they are at risk without the context of the denominator. Working in a hospital setting means residents also see a disproportionate number of positive cases and moderate to severe disease than is present in the general community.
Another common source of misinformation is the Internet and social media. Cuan-Baltazar et al. found that by February 6, 2020, of the 36 top search results for Wuhan and coronavirus, only 15 websites contained factual information (1). Educated health care workers are often assumed to be immune to online health misinformation, but social media have taken on increasing importance in the dissemination of medical information and research in health care (6). This has accelerated since the start of the pandemic, where rapidly evolving information is more quickly disseminated on social media (6,9). Of the survey respondents, 54% rated social media as their most common or very common source of medical information. Although social media are useful for rapid knowledge translation, there are risks to these non–peer-reviewed materials, including inappropriate application of context-specific resources, biased knowledge in echo chambers, insufficient source information for verification, and early adoption of unvalidated research or practice, such as the widespread prescribing of hydroxychloroquine after early social media reports of its possible activity against COVID-19 despite a lack of compelling evidence (6–8).
From early March onward, social media were flooded with reports of rising case numbers and dramatic portrayals of overwhelmed health care systems in high-burden locations such as Italy, Seattle, and New York. We therefore hypothesized that increased time spent on social media would be associated with a higher perceived risk. In our study, the median perceived risk of hospitalization was 7.5% among the low–average social media users and 17.5% among high social media users; however, this difference was not statistically different. Nor was there a significant difference between perceived risk of infection or of infecting family members, anxiety, or perceived impact of hospitalization. This held true whether or not respondents rated social media as their most common or very common source of information.
It is possible that our study was under-powered to detect a small but important difference between high and low–average social media users. However, there are other possible explanations for a lack of association. Time spent on social media may not correlate with reliance on social media for information. Given the limited information on COVID-19 in traditional learning tools, it is possible that most residents relied primarily on information from social media regardless of time spent. Our survey was also conducted several weeks into the pandemic, allowing time for discussion among residents so that messaging on social media could have spread to non–social media users.
Further limitations to our study included a small sample size and reliance on self-reported use of social media and other information sources. The response rate to our survey was also low, as expected with trainee voluntary e-mail research. It is possible that the voluntary respondents could over-represent trainees with concerns about pandemic risk. Also, respondents were from one academic institution, their average age was 28 years, and they included a high proportion of residents from internal medicine, which may limit generalizability to residents in other specialities or training at other institutions.
Conclusion
In summary, resident physicians engaged in front-line care of COVID-19 patients have a very high perceived personal COVID-19 risk that is several orders of magnitude higher than the observed incidence among the general public or health care workers. This elevated perceived risk may contribute to fear and anxiety among residents, although further study would be needed to examine how this affects resident behaviour, clinical practice, or short- and long-term mental health outcomes. Although this elevated sense of risk has several possible explanations, time spent on social media was not significantly associated with a higher perceived risk. Further study would be needed to clarify other ways in which online media influence residents’ risk perceptions and whether specific social media sites have a higher association with misinformation.
Ethics Approval:
This study was approved by the Sunnybrook Research Institute Research Ethics Board.
Informed Consent:
N/A
Registry and the Registration No. of the Study/Trial:
N/A
Funding:
No funding was received for this work.
Disclosures:
The authors have nothing to disclose.
Peer Review:
This manuscript has been peer reviewed.
Animal Studies:
N/A
References
- 1.Cuan-Baltazar JY, Muñoz-Perez MJ, Robledo-Vega C, Pérez-Zepeda MF, Soto-Vega E. Misinformation of COVID-19 on the Internet: infodemiology study. JMIR Public Health Surveill. 2020;6(2):e18444. 10.2196/18444. Medline: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Merchant RM, Lurie N. Social media and emergency preparedness in response to novel coronavirus. JAMA. 2020;323(20):2011–2012. 10.1001/jama.2020.4469. Medline: [DOI] [PubMed] [Google Scholar]
- 3.Dickmann P, Abraham T, Sarkar S, et al. Risk communication as a core public health competence in infectious disease management: development of the ECDC training curriculum and programme. Euro Surveill. 2016;21(14). 10.2807/1560-7917.ES.2016.21.14.30188. Medline: [DOI] [PubMed] [Google Scholar]
- 4.Liu M, Caputi TL, Dredze M, Kesselheim AS, Ayers JW. Internet searches for unproven COVID-19 therapies in the United States. JAMA Intern Med. 2020;180(8):1116–8. 10.1001/jamainternmed.2020.1764. Medline: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Soltaninejad K. Methanol mass poisoning outbreak: a consequence of COVID-19 pandemic and misleading messages on social media. Int J Occup Environ Med. 2020;11(3):148–150. 10.34172/ijoem.2020.1983. Medline: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Chan AKM, Nickson CP, Rudolph JW, Lee A, Joynt GM. Social media for rapid knowledge dissemination: early experience from the COVID-19 pandemic. Anaesthesia. 2020;75(12):1579–82. 10.1111/anae.15057. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Johannsson H, Selak T. Dissemination of medical publications on social media—is it the new standard? Anaesthesia. 2020;75(2):155–157. 10.1111/anae.14780. Medline: [DOI] [PubMed] [Google Scholar]
- 8.Young PJ, Nickson CP, Gantner DC. Can social media bridge the gap between research and practice? Crit Care Resusc. 2013;15(4):257–259. [PubMed] [Google Scholar]
- 9.Bhagavathula AS, Aldhaleei WA, Rahmani J, Mahabadi MA, Bandari DK. Knowledge and perceptions of COVID-19 among health care workers: cross-sectional study. JMIR Public Health Surveill. 2020;6(2):e19160. 10.2196/19160. Medline: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Burns KE, Duffett M, Kho ME, et al. A guide for the design and conduct of self-administered surveys of clinicians. CMAJ. 2008;179(3):245–252. 10.1503/cmaj.080372. Medline: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Wilson RS, Zwickle A, Walpole H. Developing a broadly applicable measure of risk perception. Risk Anal. 2019;39(4):777–91. 10.1111/risa.13207. Medline: [DOI] [PubMed] [Google Scholar]
- 12.Ontario Agency for Health Protection and Promotion (Public Health Ontario). Epidemiologic summary: COVID-19 in Ontario—January 15, 2020 to April 30, 2020. Toronto:Queen’s Printer for Ontario; 2020. [Google Scholar]
- 13.Ontario Agency for Health Protection and Promotion (Public Health Ontario). Epidemiologic summary: COVID-19 in Ontario—January 15, 2020 to October 31, 2020. Toronto:Queen’s Printer for Ontario; 2020. [Google Scholar]
- 14.Statistics Canada. Table 14-10-0023-01 labour force characteristics by industry, annual (x 1,000). Toronto: Statistics Canada. 10.25318/1410002301-eng. [DOI] [Google Scholar]
- 15.Breazzano MP, Shen J, Abdelhakim AH, et al. New York City COVID-19 resident physician exposure during exponential phase of pandemic. J Clin Invest. 2020;130(9):4726–33. 10.1172/JCI139587. Medline: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Gomez-Ochoa SA, Franco OH, Rojas LZ, et al. COVID-19 in healthcare workers: a living systematic review and meta-analysis of prevalence, risk factors, clinical characteristics and outcomes. Am J Epidemiol. 2021;190(1):161–75. 10.1093/aje/kwaa191. Medline: [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Graham-Harrison E. The whistleblower doctor who fell victim to China's coronavirus. The Guardian [Internet]. 2020. Feb 6. [cited 2020 June 6]. Available from: https://www.theguardian.com/world/2020/feb/06/li-wenliang-coronavirus-whistleblower-doctor-profile [Google Scholar]