Tracheostomies of Patients With COVID-19: A Survey of Infection Reported by Health Care Professionals

After a review of the existing literature on the performance of tracheostomy procedures during the COVID-19 pandemic and other respiratory pandemics, the research team developed an online survey for HCPs caring for patients who test positive for COVID-19. This survey was drafted on the basis of existing guidelines and literature on tracheostomies in patients who test positive for COVID-19, reviewed in fall 2020. The survey draft was shared with teams of surgeons, intensivists, and nurses in North America and South America. After incorporating feedback from all draft reviewers, the survey was submitted and approved for dissemination by the Emory University Institutional Review Board (Study 00001633) and research was carried out following the ethical standards set forth in the Helsinki Declaration. Study data were collected and managed with Research Electronic Data Capture (REDCap) tools.^34,35  Participation in the survey was voluntary, and all participants indicated their consent to be involved in the study. All participants identifying as adult HCPs were eligible for inclusion in the study. Those who did not indicate whether they had ever been infected with the SARS-CoV-2 virus were excluded from the study.

The English-language survey (Supplemental Table 1—available online only, at www.ajcconline.org) was made accessible to any visitor to the study website, www.covidtrach.org, starting November 20, 2020, and Spanish, French, and Portuguese versions were made openly accessible beginning in March 2021. The survey originally consisted of 19 questions targeted to all respondents. Those HCPs who indicated that they were involved in performing tracheostomies in patients with COVID-19 were asked 12 additional questions. These “tracheostomy providers” included a wide variety of health care team members, including those actively placing the tracheostomy tube; endoscopists; personnel monitoring patient safety; and other procedural assistants; but did not include those who cared for patients with tracheostomies without involvement in the procedure itself. Tracheostomy providers who indicated that they could describe details of how the tracheostomy procedure was performed were asked another 8 questions. Two additional questions pertaining to vaccination status were added to the survey on December 18, 2020, shortly after the beginning of COVID-19 vaccination availability for HCPs in the United States and after 25 participants had responded.

For recruitment, the survey was sent to 661 email addresses of individuals affiliated with a convenience sample of multinational hospitals, clinical academic departments, and professional organizations in critical care, pulmonology, respiratory care, and surgery (including general, trauma, thoracic, otorhinolaryngology, and oral and maxillofacial surgery; Supplemental Table 2—available online only). Social media recruitment included dissemination via moderated feeds and paid Facebook advertising.

Survey responses were exported from REDCap for analysis, and surveys in which the respondent did not indicate completion of the survey by advancing through all questions were excluded. However, if the respondent advanced through the survey but skipped questions, these were included. Descriptive analysis was performed for providers reporting a history of COVID-19 infection (“infected”) and those who reported no history of infection (“uninfected”). All categorical demographic and COVID-19–related data were compared using χ² or Fisher exact tests. Significance was set to α = .05 (ie, a 95% CI).

Responses submitted before February 1, 2021, were separated from those entered on or after February 1 to assess the effect of pandemic time course. By this date, all North American and Western European countries, as well as the majority of South American and Asian countries, had made COVID-19 vaccines available to health care providers.³⁶  To assess COVID-19 pandemic severity, cumulative confirmed COVID-19 cases per million people were matched to survey respondents by response date and country of employment.¹  To assess country income, country of employment was linked to World Bank country income status.³⁷  High-income countries (HICs) were compared with low-income and middle-income countries (LMICs). Five-point Likert-scale responses were collapsed into 3 categories: “not at all” or “mildly”, “somewhat”, and “very” or “extremely.”

Providers who reported having been involved in performing tracheostomies on patients who tested positive for COVID-19, including bronchoscopists, physicians, nurses, advanced practice providers, respiratory care practitioners, and other technicians working in critical care units involved in the procedure, were compared with those who reported that they had not been involved in performing tracheostomies on patients who tested positive for COVID-19. Multivariable logistic regression assessing reported COVID-19 infection as the outcome was performed with 3 variables selected a priori (tracheostomy provider status, COVID-19 prevalence, and pandemic timeline). Because of the large number of potentially clinically relevant variables, as well as the rapid flux in knowledge of the pertinence of those risk factors, we decided to consider all variables with P < .20 for potential inclusion. We then excluded colinear variables and those not clinically relevant.

Variables considered for inclusion in the adjusted model included country and country income level, gender, race, age, HCP type, years in practice, frequency of caring for patients who tested positive for COVID-19, vaccination status, concern for contracting COVID-19 infection, and facility type.^2,4,38–44  Manual backward stepwise elimination was then used for variable selection. Because of complete separation with pandemic time course and country income level, individual country was excluded. The variables thus selected for the final adjusted model were tracheostomy provider status, country income level, pandemic timeline, and country’s COVID-19 caseload.

Separate descriptive analyses were done regarding tracheostomy-specific factors reported by providers who described the procedure in detail. All statistical analyses were performed with SAS version 9.4 by Emory University’s Biostatistics Collaboration Core staff.

The survey link on the study website, www.covidtrach.org, was accessed from 370 unique inter-net protocol (IP) addresses according to the website service Squarespace. Individual single-use survey links were also shared directly with 661 email addresses. Unique IP addresses and single-use survey links were used to minimize the possibility of duplicate survey responses.

Of the 1031 people who received an email or clicked on the survey on the public website, 365 people completed the survey, a 35% completion rate. (Completion was defined as advancing through the entire survey, even if some questions were skipped.) Four of the participants who completed the survey did not respond to the question regarding history of COVID-19 infection and were excluded. Thus, 361 complete survey responses were included in the analysis.

The HCPs who responded to the survey represented 33 countries, with large proportions working in India (32%) and the United States (41%; Table 1 and Supplemental Table 3—available online only). Seventy-three percent of all HCPs responded to the survey after February 1, 2021, and 44% worked in LMICs. COVID-19 infection was significantly more frequently reported by HCPs who were based outside the United States (P < .001); in LMICs (P < .001); non-White (P = .02); younger than 45 years (P = .01); and responding February 1, 2021, or later during pandemic (P = .005; Table 1). Nurses, respiratory care practitioners, technicians, and therapists (P = .005) and those caring for patients who tested positive for COVID-19 at least weekly (P = .03) were also significantly more likely to report COVID-19 infection than other types of HCPs or those working less often with patients who tested positive for COVID-19 (Table 2).

The median cumulative number of confirmed cases of COVID-19 per million people by date of survey response and country (date-location) was 46 326 (range, 27 to 157 103).¹  Those HCPs responding at a date-location with a COVID-19 caseload below the median reported COVID-19 infection more frequently than did those responding from a date-location with a caseload above the median (P < .001, Table 1). Ninety-six percent of survey responses from LMICs, including 100% of responses from India, were associated with below-median caseloads.

Ninety percent of HCPs were partially or fully vaccinated at the time of survey response (Table 2), excluding the 25 individuals who responded to the survey before the vaccine-related questions were added on December 18, 2020. Among vaccinated HCPs who reported COVID-19 infection and indicated timing relative to vaccination, 69% indicated that they were infected before vaccination, with the remaining 31% becoming infected after partial or full vaccination. Thirteen of the vaccinated HCPs who reported COVID-19 infection did not indicate timing relative to vaccination status. The 4 HCPs who reported COVID-19 infection and responded to the survey before the vaccine-related questions were added were excluded from vaccine analysis.

Approximately half of respondents (n = 179) indicated involvement in performing tracheostomies on patients who tested positive for COVID-19 (Table 2). COVID-19 tracheostomy providers represented 58% of the infected HCP population and 47% of the uninfected population (P = .06). Of the 53 infected tracheostomy providers, 76% reported infection more than 14 days after involvement in a tracheostomy on a patient who tested positive for COVID-19, whereas 6% reported infection within 14 days of a tracheostomy. Eighteen percent were unsure of timing; 2 providers did not respond to the question (Supplemental Table 4—available online only). Infection rates reported according to tracheostomy provider status and vaccination status were also examined and demonstrated that unvaccinated HCPs who were not involved with any tracheostomy procedures accounted for 22% of those reporting infection but only 8% of those not reporting infection (P = .02; Supplemental Table 5— available online only). This pattern was not observed among tracheostomy providers, with unvaccinated individuals accounting for similar proportions of infected and uninfected respondents (P = .94).

COVID-19 tracheostomy providers were not more likely to report infection in the multivariable model (odds ratio [OR], 1.48; 95% CI, 0.90–2.46; P = .13; see Figure). However, after tracheostomy provider status, pandemic timeline, and COVID-19 caseload were adjusted for, HCPs from LMICs were significantly more likely to report a history of COVID-19 infection (OR, 2.88; 95% CI, 1.50-5.53; P = .001).

Descriptive data of the LMIC and HIC samples are available in Supplemental Table 6 (available online only). The LMIC population was majority non-White (87% vs 30% of the HIC population); younger than 45 years old (LMIC 90% vs HIC 70%); and nurses, respiratory care practitioners, technicians, or therapists (LMIC 73% vs HIC 31%). Moreover, 13% of LMIC respondents were unvaccinated, compared with 6% of HIC respondents, and 23% of LMIC respondents indicated that less than 75% of their needs for PPE were met, whereas only 8% of HIC respondents indicated this concern.

A subset analysis of the COVID-19 tracheostomy providers was performed to identify procedures that may have contributed to infection prevention. Of the 179 COVID-19 tracheostomy providers, 145 (81%) were able to describe procedural details and were included in the subset analysis. Demographics of this population are detailed in Supplemental Table 7 (available online only).

Most tracheostomies within this dataset were performed in an intensive care unit (81%), percutaneously (69%), and within 14 days of a patient’s COVID-19 diagnosis (56%) (Table 3). Most COVID-19 tracheostomy providers also used some form of aerosol-reducing precautions (83%). These precautions included minimizing bronchoscopy, discontinuing mechanical ventilation, and using additional barriers. Major complication rates were relatively low, with the most frequently reported complications being minor bleeding and significant desaturation (Supplemental Table 8—available online only). Given that performing tracheostomies on patients who tested positive for COVID-19 was not associated with increased reported COVID-19 infection in the provider and that most infections reported by tracheostomy providers were identified more than 14 days after the procedure, statistical comparison of procedural factors was not performed.

Minimizing HCP infection risk while maintaining an active workforce to care for patients has been critical to health care system stability throughout the COVID-19 pandemic.⁴⁵  Although tracheostomies are aerosol-generating procedures with theoretical risks to providers, these risks must be balanced against optimizing patient care and resource utilization during a pandemic.⁶  This study indicates that providers who are involved in performing tracheostomies on patients who test positive for COVID-19 do not report greater rates of infection than other HCPs. However, the data suggest that HCPs working in LMICs are more likely to report COVID-19 infection than their counterparts who work in HICs, after adjusting for country COVID-19 caseload, pandemic timeline, and tracheostomy provider status.

Despite potential risks delineated in many tracheostomy guidelines, the lack of association between involvement in tracheostomies and reported COVID-19 infection is consistent with existing literature.^{15,21,25,30,31,33,46–48}  In a recent review of tracheostomy outcomes during the pandemic, only 3 of the 58 papers examined identified HCP infection within 4 weeks following tracheostomy involvement.³³  Although 1 study reported a COVID-19 infection rate of 7.7% among HCPs involved in tracheostomies, this rate did not significantly differ from the 11.5% overall HCP infection rate reported.⁴⁷  Nonetheless, several expert groups have recommended delaying tracheostomy or deferring procedures entirely on patients who test positive for COVID-19.^27–31  Such recommendations, although aiming to protect HCPs during a time of uncertainty, may have prevented optimal patient care or use of resources.

In addition to the absence of statistical association between involvement in tracheostomies and reported COVID-19 infection generally, only 3 tracheostomy providers reported being diagnosed with COVID-19 within 14 days of performing a tracheostomy on an infected patient. This finding is particularly notable considering that more than 50% of tracheostomy providers reported performing the procedure within 14 days of when the patient had been diagnosed with COVID-19. This frequency of such “early” tracheostomies is perhaps surprising given that more than 90% of protocols from the first several months of the pandemic suggested delaying the procedure until after 14 days of intubation.⁴⁹  Early tracheostomy may have multiple benefits for patients, including reduced ventilator dependence,⁵⁰  shorter stay in the intensive care unit,⁵¹  decreased analgesia requirements,⁵²  and faster return to physical activity^53,54 ; it is also cost-effective.^55,56  This suggests that, in appropriately selected patients, early tracheostomy may assist with both patient care and resource management during a crisis.⁵⁷ 

Collectively, the data imply that tracheostomies can be safely performed on patients who test positive for COVID-19 with appropriate precautions and at the appropriate time. It must be noted that nearly all (98%) tracheostomy providers reported wearing particulate-filtering facepiece respirators during tracheostomies. Moreover, 83% reported using at least 1 aerosol-reducing maneuver, including discontinuation of mechanical ventilation, use of additional barriers, and minimizing bronchoscopy. As suggested above, tracheostomy timing may also be a factor affecting reported infection rates. Even patients undergoing “early” tracheostomy within 14 days of diagnosis or 10 days of intubation have typically developed antiviral antibodies and infectivity is much diminished.²⁶  Data were not collected regarding the use of specific facility-based protocols for tracheostomy timing or methods, but such guidelines may have also been protective. Further research should evaluate the specific factors that optimize safety of tracheostomy providers in greater depth.

The findings from this survey have important implications for HCP safety irrespective of involvement in tracheostomies. The 25% COVID-19 infection rate reported (Table 1) is consistent with existing data demonstrating that HCPs are at increased risk of infection relative to the general population.^2,3  Existing data point to multiple factors contributing to infection risk among HCPs, including use of personal protective equipment (PPE) , provider role, and race.^{2,5,43,44,58,59}  Similarly, this survey demonstrates associations between younger age, increased frequency of COVID-19 patient care, non-White race, and HCP role (nurse, respiratory care practitioner, technician, or therapist) with increased reported COVID-19 infection in univariable analysis. The data also highlight a protective role of COVID-19 vaccination among HCPs who are not involved with tracheostomies. Neither the fact that these additional factors, such as HCP role, were nonsignificant in multi-variable analysis nor the lack of a significant protective effect of vaccination among tracheostomy providers indicates that these factors are unimportant for infection risk. The present study was neither designed nor powered to examine these factors fully, and additional research is needed to understand these factors’ contribution to COVID-19 risk among HCPs.

In univariable analysis, COVID-19 infection was also reported more often by HCPs located in LMICs, by those who responded to the survey later in the pandemic, and, counterintuitively, by HCPs from countries with below-median cumulative confirmed COVID-19 cases per million people. After adjusting for country caseload and pandemic timeline in multivariable analysis, LMIC status remained significantly associated with increased reported COVID-19 infection. Others have reported significant disparities in HCP safety at a global level, including inadequate access to PPE, reduced use of safety protocols, and disparity in vaccine access.^60,61  Given that this study was not designed to assess differences between respondents in LMICs and those in HICs, inferential statistical tests were not performed. However, descriptive analysis of the study sample by income status highlights the many differences between the populations, including age, race, vaccination status, and PPE availability. Further research designed to examine these and other factors that could contribute to increased HCP infection in LMICs is needed.

The counterintuitive finding in univariable analysis that lower COVID-19 caseload was linked to increased reported infection may be due to this variable’s relationship with country income level. HCPs in LMICs accounted for 84% of the surveys associated with a caseload below the median and only 4% of surveys associated with a caseload above the median. COVID-19 cases may be dramatically underreported in LMICs such as India and elsewhere.^62–64  Thus, the impact of COVID-19 prevalence on HCP infection rates should not be dismissed. However, the data presented here indicate that additional country-specific factors play an important role in infection risk. The risks of working in a LMIC may outweigh the risks associated with high COVID-19 prevalence. These findings are consistent with existing literature providing evidence of significant disparities in HCP safety at a global level.⁶⁰ 

This study has several limitations. All associations are correlative, and cause should not be inferred. The HCP COVID-19 infection rates are based on self-report without submission of laboratory testing or other confirmation of infection and are thus subject to recall bias. However, the frequency of COVID-19 infections reported in our study, both among all HCPs and among tracheostomy providers specifically, resembles that reported elsewhere.^2,33,59 

Selection bias is likely both in the targeted convenience sampling method used and in the self-selection of survey respondents. Recruitment methods involved highlighting the nature of the study and its relevance to tracheostomy care; the fact that nearly 50% of participants identified as being involved in tracheostomy care reflects these methods. Moreover, though data to this effect were not collected, many respondents not involved in the index procedure were most likely still involved in caring for patients with tracheostomies given the recruitment methods used. Given the publicly available nature of the survey and distribution at the organizational level in some cases, we are unable to estimate the full reach of study and thus we could not arrive at a conclusive response rate. The calculated survey completion rate (defined as those who completed the survey accessed either via email or via a public website) was 35% despite the wide distribution of the survey, which is similar to that reported from other surveys of HCPs on sensitive topics.^65,66  Nonetheless, the results should be interpreted cautiously given that the responses may not be generalizable to the complete population of HCPs worldwide.

Selection bias also most likely contributed to the geographic distribution of survey respondents. Although 33 countries are represented, 74% of the respondents worked in either the United States or India. Providers and facilities in the United States and India were heavily targeted during recruitment, given they are 2 of the countries with the greatest burden of patients who test positive for COVID-19 worldwide. However, these 2 countries accounted for only 34% of the global COVID-19 case burden as of July 16, 2021 (the date of survey closure).¹  This overrepresentation means that findings with respect to country income may be due to differences between the United States and India as opposed to differences between HICs and LMICs generally. Moreover, the use of the categories of HIC and LMIC as defined by the World Bank is also imperfect, as substantial variation in income level exists within countries. Further studies with greater representation from other HICs and LMICs, as well as more detailed analysis of regional income levels, are needed.

The longer time frame during which the survey was made available is both a strength and a weakness of the study. The 7-month time frame allowed us to target different countries at different times to account for global pandemic fluctuations. Non-English versions of the survey were distributed later during the study as well, corresponding to these fluctuations. This design, while intended to optimize response rates from HCPs at date-locations most affected by the pandemic, makes it difficult to evaluate the interrelationships between geography, time, language, and COVID-19 prevalence. Additional epidemiologic research will be necessary to understand the fluctuations in risk factors throughout the pandemic.

This multinational survey demonstrates that tracheostomy providers do not report greater COVID-19 infection than other HCPs. This suggests that, if adequate safety measures are in place, tracheostomies can be performed on patients who test positive for COVID-19 without putting HCPs at increased infection risk. This study should inform provider decision-making as COVID-19 shifts from a pandemic to an endemic problem. After adjusting for tracheostomy provider status, country caseload, and pandemic time-line, LMIC providers reported higher COVID-19 infection rates. Although overrepresentation of the United States and India may contribute to this finding, the results highlight global health care disparities and the impact of resource limitations. In conclusion, HCPs should continue to perform tracheostomies in patients who test positive for COVID-19 as indicated by standards of care and using adequate safety measures. However, additional research on infection risk in lower-resource settings is warranted.

PTS-COVIDTrach Collaborative authors include Kerianne Allen, BA, CCRP, Riverside Health System, Newport News, Virginia; Marissa Boeck, MD, MPH, University of California, San Francisco; Christine A. Castater, MD, Emory University and Grady Memorial Hospital, Atlanta, Georgia; Michael Farrell, MD, MS, University of California, San Francisco; Vikas Pathak, MD, ATSF, DAABIP, Riverside Health System, Newport News, Virginia; Keriann Marie Van Nostrand, MD, Emory University, Atlanta, Georgia and University of South Florida, Tampa, Florida.

We are grateful for the assistance of colleagues in the Indian Association of Respiratory Care and the Multinational Council for Respiratory Care and our volunteer translators, including Daniela Farchi, Bahaa Kazzi, Eliana Lillivek, and Gustavo Moraes. We would also like to acknowledge funding support from the Emory University School of Medicine “Imagine, Innovate, Inspire” Nexus Award and Library Information Technology Services grant (UL1 TR000424) for REDCap.

Finally, we would like to acknowledge the support from all members of the PTS-COVIDTrach Collaborative: Katrina Abril, MD, MSc, Emory University, Atlanta, Georgia, and Spectrum Health System, Grand Rapids, Michigan; Sérgio S. Arap, MD, PhD, Hospital Das Clínicas, University of São Paulo, Brazil; Cindy L. Austin, MS, CCRP, Mercy Hospital, Springfield, Missouri; Mark Barry, MD, University of California, San Francisco; Ricardo F. Bento, MD, PhD, Hospital Das Clínicas, University of São Paulo, Brazil; Juan Blas, MD, MBBS-MBChB, Royo Villanova Hospital, Zaragosa, Spain; Maraya Camazine, MS, University of Missouri School of Medicine; Pedro P.D. Ciaralo, MD, Hospital Das Clínicas, University of São Paulo, Brazil; Jeffrey Coughenour, MD, University of Missouri School of Medicine; Deidre Dillon, BS, University of Missouri School of Medicine; Anya Greenberg, MBA, University of California, San Francisco; Rui Imamura, MD, PhD, Royo Villanova Hospital, Zaragosa, Spain; Wissam S. Jaber, MD, Emory University, Atlanta, Georgia; Or Kalchiem-Dekel, MD, Memorial Sloan Kettering Cancer Center, New York, New York; Hannah Kim, BS, ATC, University of California, San Francisco; Luiz P. Kowalski, MD, PhD, Hospital Das Clínicas, University of São Paulo, Brazil; Ali Mahmoud, MD, Hospital Das Clínicas, University of São Paulo, Brazil; Alessandro W. Mariani, MD, PhD, Hospital Das Clínicas, University of São Paulo, Brazil; Carlos A.M. Menegozzo, MD, Hospital Das Clínicas, University of São Paulo, Brazil; Phoebe Miller, MS, University of California, San Francisco; Hélio Minamoto, MD, PhD, Hospital Das Clínicas, University of São Paulo, Brazil; Fábio L.M. Montenegro, MD, PhD, Hospital Das Clínicas, University of São Paulo, Brazil; Jonathan Nguyen, DO, Grady Memorial Hospital and Morehouse University, Atlanta, Georgia; Paul Park, MS, University of California, San Francisco; Paulo M. Pêgo-Fernandes, MD, PhD, Hospital Das Clínicas, University of São Paulo, Brazil; Jennifer Randolph, BSN, RN, University of Missouri School of Medicine; Manu Sancheti, MD, Emory University, Atlanta, Georgia; Jones Santos Jr, MD, Hospital Das Clínicas, University of São Paulo, Brazil; Alejandro Sardi, MD, Emory University, Atlanta, Georgia; Matthew Schimmel, MD, Emory University, Atlanta, Georgia; Issa Talal, PhD, MSc, MD, MBBS-MBChB, Royo Villanova Hospital, Zaragosa, Spain; Gabriel Tirado, MD, MBBS-MBChB, Royo Villanova Hospital, Zaragosa, Spain; Edivaldo M. Utiyama, MD, PhD, Hospital Das Clínicas, University of São Paulo, Brazil; Thomas G. Weiser, MD, MPH, Stanford University, Palo Alto, California; Cheryl White, BSN, RN, Riverside Health System, Newport News, Virginia; Carlos Yánez Benitez, MD, MSc, Royo Villanova Hospital, Zaragosa, Spain.