Late tracheostomy was associated with a 16% decreased chance in discontinuing mechanical ventilation (hazard ratio [HR], 0.84; 95% CI, 0.55-1.28; P = .41). Tracheostomy timing was not associated with chance of death (HR, 1.03; 95% CI, 0.47-2.25; P = .94).
eAppendix. NYULH indications and guidelines used in the selection of patients for early tracheostomy during the SARS-CoV-2 pandemic.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Kwak PE, Connors JR, Benedict PA, et al. Early Outcomes From Early Tracheostomy for Patients With COVID-19. JAMA Otolaryngol Head Neck Surg. 2021;147(3):239–244. doi:10.1001/jamaoto.2020.4837
Is early tracheostomy associated with improved outcomes for patients with coronavirus disease 2019 (COVID-19) requiring mechanical ventilation?
In this cohort study of 148 patients with COVID-19, timing of tracheostomy was significantly associated with length of stay; median length of stay was 40 days in those who underwent early tracheostomy and 49 days in those who underwent late tracheostomy.
In patients with COVID-19, early tracheostomy was noninferior to late tracheostomy and may be associated with improvement in some outcomes; it did not contribute to increased infections of clinicians.
Decision-making in the timing of tracheostomy in patients with coronavirus disease 2019 (COVID-19) has centered on the intersection of long-standing debates on the benefits of early vs late tracheostomy, assumptions about timelines of infectivity of the novel coronavirus, and concern over risk to surgeons performing tracheostomy. Multiple consensus guidelines recommend avoiding or delaying tracheostomy, without evidence to indicate anticipated improvement in outcomes as a result.
To assess outcomes from early tracheostomy in the airway management of patients with COVID-19 requiring mechanical ventilation.
Design, Setting, and Participants
A retrospective medical record review was completed of 148 patients with reverse transcriptase–polymerase chain reaction–confirmed COVID-19 requiring mechanical ventilation at a single tertiary-care medical center in New York City from March 1 to May 7, 2020.
Open or percutaneous tracheostomy.
Main Outcomes and Measures
The primary outcomes were time from symptom onset to (1) endotracheal intubation, (2) tracheostomy; time from endotracheal intubation to tracheostomy; time from tracheostomy to (1) tracheostomy tube downsizing, (2) decannulation; total time on mechanical ventilation; and total length of stay.
Participants included 148 patients, 120 men and 28 women, with an overall mean (SD) age of 58.1 (15.8) years. Mean (SD; median) time from symptom onset to intubation was 10.57 (6.58; 9) days; from symptom onset to tracheostomy, 22.76 (8.84; 21) days; and from endotracheal intubation to tracheostomy, 12.23 (6.82; 12) days. The mean (SD; median) time to discontinuation of mechanical ventilation was 33.49 (18.82; 27) days; from tracheostomy to first downsize, 23.02 (13.76; 19) days; and from tracheostomy to decannulation, 30.16 (16.00; 26) days. The mean (SD; median) length of stay for all patients was 51.29 (23.66; 45) days. Timing of tracheostomy was significantly associated with length of stay: median length of stay was 40 days in those who underwent early tracheostomy (within 10 days of endotracheal intubation) and 49 days in those who underwent late tracheostomy (median difference, −8; 95% CI, −15 to −1). In a competing risks model with death as the competing risk, the late tracheostomy group was 16% less likely to discontinue mechanical ventilation (hazard ratio, 0.84; 95% CI, 0.55 to 1.28).
Conclusions and Relevance
This cohort study from the first 2 months of the pandemic in New York City provides an opportunity to reconsider guidelines for tracheostomy for patients with COVID-19. Findings demonstrated noninferiority of early tracheostomy and challenges recommendations to categorically delay or avoid tracheostomy in this patient population. When aligned with emerging evidence about the timeline of infectivity of the novel coronavirus, this approach may optimize outcomes from tracheostomy while keeping clinicians safe.
The novel coronavirus, severe acute respiratory syndrome (SARS) coronavirus 2 (SARS-CoV-2), ravaged New York City in early March 2020, precipitating an exponential escalation of new cases within days and rendering the city the US epicenter of this pandemic. Amid the early proliferation of confounding and severe clinical presentations, which outpaced systematic data collection, one of the chief quandaries for airway surgeons managing this respiratory disease centered on the appropriate timing and management of tracheostomy. Within the first weeks of the pandemic, at least 7 sets of guidelines emerged from academic journals in otolaryngology, anesthesia, and thoracic surgery, rooted in expert consensus and opinion. Most recommended delaying or avoiding tracheostomy to minimize risk of infection to clinicians and because the pulmonary manifestations and clinical trajectory of the disease, and therefore the anticipated benefit of tracheostomy, were unclear.1-7
These consensus guidelines synthesized principles from ongoing debates about early vs late tracheostomy with experience from the previous SARS epidemic centered in China, with the presumption that the novel SARS-CoV-2 might mirror the clinical trajectory of its predecessor.8 Specifically, recommendations for tracheostomy in the current pandemic were rooted in the assumption that maximal infectivity of this novel virus occurred around day 7 to 10 after symptom onset, so that performing tracheostomy at that time would engender maximal risk to those performing it.9
Based on previously established principles of airway and tracheostomy management, a multidisciplinary team of surgeons at this institution began, from the outset of the pandemic, to identify patients who would benefit from earlier tracheostomy. Patients who met criteria underwent tracheostomy and were followed longitudinally by this multidisciplinary team from their admission until the defined end point of the study. This report includes outcomes from the first 2 months of this operational protocol at this hospital.
This study was reviewed and approved by the institutional review board at NYU Langone Health, waiving patient informed consent owing to deidentified data. A retrospective medical record review was performed to identify all patients who tested positive for SARS-CoV-2 on real-time reverse transcriptase–polymerase chain reaction assay testing who were intubated and subsequently underwent percutaneous or open tracheostomy at our institution between March 1 and May 7, 2020. The polymerase chain reaction assay chosen for diagnosis was that which was immediately available at NYU Langone Health at the time of testing and included Cobas (Roche Diagnostics) and Xpert Xpress (Cepheid) SARS-CoV-2 assays. This was performed by nasopharyngeal swab at the time of the patient’s arrival to the emergency department.
The care of these patients was directed primarily by intensivists, including interventional pulmonologists who performed percutaneous tracheostomies. Percutaneous tracheostomy was performed at the bedside in the intensive care unit by a novel technique reported earlier by this group.10 The Otolaryngology–Head and Neck Surgery service was consulted to perform tracheostomy on patients with difficult anatomy or if, owing to massive volume of patients, the interventional pulmonologists were not available. Open tracheostomy was also performed at the bedside in the intensive care unit with standard surgical technique.
Patient demographic information included age, sex, race/ethnicity, body mass index (calculated as weight in kilograms divided by height in meters squared), and Charlson Comorbidity Index score.11,12 The following parameters were used to assess the hospital course of these patients: (1) time from symptom onset, as documented by patient report in the admission history and physical, to (a) intubation, (b) tracheostomy; (2) time from intubation to (a) tracheostomy, (b) discontinuation of mechanical ventilation; and (3) time from tracheostomy to (a) downsizing of tracheostomy tube, (b) decannulation. Total length of stay was calculated.
The timing of the tracheostomy procedure was subcategorized as either early or late tracheostomy, with early tracheostomy defined as procedures performed prior to day 10 of intubation and late tracheostomy defined as procedures performed at day 10 of intubation or later (for listed indications and guidelines used for selecting patients for early tracheostomy, please see the eAppendix in the Supplement). Definitions of early and late tracheostomy have varied across reviews and meta-analyses; 10 days was chosen for this analysis as the point of demarcation as an average across these reviews and meta-analyses.13-15 All patients who underwent early tracheostomy met all criteria in the eAppendix in the Supplement.
Time to discontinuation of mechanical ventilation was measured from the day of intubation to the first day a patient was able to tolerate tracheostomy collar for 24 continuous hours. Time to downsize tracheostomy tube was measured from the day of tracheostomy to the day of first tube exchange from a cuffed to cuffless tracheostomy tube (or in a small minority of patients, the change from a larger to smaller size of cuffed tube). Time to decannulation was measured from the day of tracheostomy to the day of decannulation. Length of stay was measured from the day of admission to the day of discharge. A distinction was not made between discharge to home, to inpatient or outpatient rehabilitation, or to long-term acute care hospitals. Major in-hospital events were noted, including the need for extracorporeal membrane oxygenation (ECMO), cardiac arrest, and cerebrovascular accident (CVA).
Descriptive statistics were used to summarize variables with frequencies and percentages for categorical variables or means and SDs for continuous variables. The median associated with the interquartile range (IQR) was used as well if continuous variables were skewed. The associations between categorical variables were evaluated with the Fisher exact test. For continuous variables, the Kolmogorov-Smirnov normality test was used to evaluate the distribution of the following hospital course variables: time to intubation, time to tracheostomy from symptom onset, time to tracheostomy from intubation, time to discontinuation of mechanical ventilation, time to downsize tracheostomy tube, time to decannulation, and length of stay. Univariate analyses of the association between patient demographic characteristics and hospital course and of the association between major in-hospital events and hospital course were performed using a 2-sample t test in cases of normal population distribution. If the normality test rejected the null hypothesis, a Wilcoxon rank-sum test was used to compare patient demographic characteristics and in-hospital events. The differences of hospital course variables between tracheostomy timing groups were quantified by the mean or median associated with its 95% CI. To compare the cumulative incidence function on the time to discontinuation of mechanical ventilation between early and late tracheostomy, the Fine and Gray’s competing risk model was used to quantify the difference in the hazard ratio (HR) by treating death as a competing risk, while controlling for potential confounding variables.16 All statistical analyses were performed using R, version 4.0.2 statistical software (RStudio 1.3.1056, R Foundation for Statistical Computing). All P values reported were 2-sided; statistical significance was defined as P < .05.
From March 1 to May 7, 2020, 148 tracheostomies were performed for critically ill patients with SARS-CoV-2. These patients were predominantly male (120; 81%) and obese (mean [SD] body mass index, 30.57 [6.48]) with a mean (SD) age of 58.1 (15.8) years. The comorbidities with the highest prevalence were hypertension (80 [54%]), diabetes (47 [32%]), and ischemic heart disease (20 [14%]). The mean (SD) Charlson Comorbidity Index score was 2.05 (1.58). Twenty-nine patients (20%) required ECMO, 11 patients (7%) had a documented cardiac arrest, and 11 patients (7%) had a documented CVA during their hospital course. Table 1 summarizes patient demographic characteristics, comorbidities, and major in-hospital events.
The mean (SD) time to intubation from symptom onset was 10.57 (6.58) (median [IQR], 9 ) days. The mean (SD) time to tracheostomy was 22.76 (8.84) (median [IQR], 21[11.5]) days from symptom onset and 12.23 (6.82) (median [IQR], 12 ) days from intubation. For the early tracheostomy group, the mean (SD) time to tracheostomy was 17.38 (7.27) (median [IQR], 16 [5.75]) days from symptom onset and 5.58 (2.12) (median [IQR], 6 ) days from intubation. For the late tracheostomy group, the mean (SD) time to tracheostomy was 25.69 (8.24) (median [IQR], 24 ) days from symptom onset and 15.83 (5.67) (median [IQR], 14 ) days from intubation. Fifty-two patients (35%) underwent early tracheostomy. The mean (SD) time to discontinuation of ventilation was 33.49 (18.82) (median [IQR], 27 [18.5]) days from intubation, time to first downsize was 23.02 (13.76) (median [IQR], 19 [13.5]) days from tracheostomy, time to decannulation was 30.16 (16.00) (median [IQR], 26 ) days from tracheostomy, and length of stay was 51.29 (23.66) (median [IQR], 45 [27.7]) days (Table 2).
At the completion of the study period, 108 patients (73%) had discontinued mechanical ventilation, 94 (64%) had been decannulated, 107 (72%) had been discharged, 54 (36%) had documented stoma closure, and 30 (20%) had died (Figure 1). Ten patients (7%) remained mechanically ventilated without a specified disposition at the completion of the study period. All hospital course variables failed the null hypothesis of normal distribution. When looking at the timing of tracheostomy, the median time to discontinuation of mechanical ventilation from intubation was 26.5 days for the early tracheostomy group and 31 days for the late tracheostomy group (median difference, −4; 95% CI, −9 to 2). The median time to decannulation was 27.5 days for the early group and 24 days for the late group (median difference, 3; 95% CI, −2 to 8). The median length of stay was 40 days for the early group and 49 days for the late group (median difference, −8; 95% CI, −15 to −1) (Table 2).
A competing risk model was used to compare discontinuation of mechanical ventilation vs death in patients with early and late tracheostomies (Figure 2). On univariate analysis, the early tracheostomy group was 16.3% more likely to discontinue mechanical ventilation (HR, 0.84; 95% CI, 0.55 to 1.28). A competing risks model accounting for age, sex, Charlson Comorbidity Index score, in-hospital events (ECMO, CVA), and early vs late tracheostomy did not show a significant association between tracheostomy timing and discontinuation of mechanical ventilation.
There were no complications intraoperatively or intraprocedurally in any of the patients in this cohort. However, 10 patients experienced postoperative bleeding, which, in all cases, was controlled at the bedside without requiring operative intervention. Two patients experienced peristomal soft tissue necrosis, which was managed expectantly and with diligent wound care; both were decannulated and did not manifest a persistent tracheocutaneous fistula.
At this institution, tracheostomy for patients with COVID-19 was performed by the Interventional Pulmonology or Otolaryngology service. Of 3 pulmonologists who performed tracheostomy, none contracted COVID-19. The Department of Otolaryngology–Head and Neck Surgery faculty totals 35 people; of these, 6 contracted COVID-19. Of these, 5 were never involved in performing tracheostomies before they fell ill; 1 surgeon involved in tracheostomies contracted the virus. All have recovered completely.
This study aggregates retrospective data from our institution in the US epicenter of the COVID-19 pandemic, where our protocol for timing of tracheostomy represented an alternate approach to guidelines that emerged in the early days of the pandemic from consensus expert panels. It seemed that these guidelines emerged from 3 interweaving strands of ongoing debate about tracheostomy and the very nature of COVID-19 itself: (1) safety to health care clinicians performing and caring for tracheostomies, (2) the appropriate timing of tracheostomy, and (3) whether outcomes, pulmonary or otherwise, would be influenced by tracheostomy.
Questions about safety to clinicians performing tracheostomy grew largely from the proliferation of the term super spreading event, verbiage used by the World Health Organization during the 2003 pandemic and repeated by many to describe events that were thought likely to increase the potential for transmission of live virus to clinicians.17 Guidelines from the Centers for Disease Control and Prevention, written in an effort to reduce risk to clinicians during the 2003 pandemic, were reproduced and promulgated at the start of the current pandemic. In the early days of the COVID-19 pandemic, it was presumed that SARS coronavirus would logically be a reasonable template for modeling practices to address SARS-CoV-2. Tracheostomy, then, was grouped with endotracheal intubation and flexible laryngoscopy, as an aerosol-generating procedure.
Characterizing tracheostomy this way led many to create a guideline that failed to account for the timing of tracheostomy in the context of the patient’s entire hospital course and symptom onset; our data illustrate the value of this timeline context. One of the most important data points in this analysis is the mean number of days from symptom onset to tracheostomy, which was 22.76—more than 3 weeks from symptom onset. In addition, we agreed on 10 days as the point of demarcation between early and late tracheostomy, synthesizing many previous reports and reviews of early vs late tracheostomy; still, we found that the mean time from intubation to tracheostomy in this cohort was 12.23 days, despite efforts to perform tracheostomy during an early time frame, within the selection algorithm listed in the eAppendix in the Supplement.
Emerging evidence about viral load and infectivity of the novel coronavirus further demonstrates the importance of these data points and calls into question the notion of purposefully delaying or avoiding tracheostomy in this patient cohort. The original SARS coronavirus manifested peak infectivity at 9 to 10 days following symptom onset; tracheostomy around that time would therefore be risky to those performing it. However, the novel coronavirus seems to demonstrate peak infectivity at, or even 2 days prior to, symptom onset, and steadily decreasing infectivity thereafter.18-21 These initial studies were carried out in small cohorts of patients with mild disease, so it is conceivable that patients with severe manifestations of disease have higher viral loads that follow a similar trajectory; nonetheless, by the time a tracheostomy is performed, at a mean of 22 days from symptom onset, the infectivity of tracheal secretions is likely to be minimal.
Synthesizing the virology of the novel coronavirus, as well as situating this information in the context of the full natural history of this disease, demonstrates the false dichotomy of early vs late tracheostomy, particularly in the management of COVID-19. With regard to the question of safety of tracheostomy to surgeons, data from the present study suggest that the number of days that have elapsed since intubation may be less relevant than the number of days since the patient first began experiencing symptoms.
Furthermore, these data demonstrate noninferiority in performing tracheostomy earlier in the clinical course of thoughtfully selected patients. In patients who underwent early tracheostomy, length of stay was shorter than in those who underwent late tracheostomy. We postulate that the vast range of severity in clinical symptoms, which has come to be a hallmark of this disease, obfuscates neat categorization of patients into early and late tracheostomy groups and undermines the historical significance of that distinction. This analysis points toward a more multifaceted decision-making process about timing of tracheostomy in patients with COVID-19; the simple number of days that elapse since intubation is but 1 consideration for clinicians among a complex set of data points that may include indicators of disease severity, emerging translational evidence about viral load and infectivity across time, and nonsurgical interventions and therapies that may alter the clinical course of patients with COVID-19 with severe disease.
Ultimately, this approach was rooted in a commitment to proactive patient care and a hypothesis borne out of experience that earlier tracheostomy would be associated with better outcomes for a methodically selected cohort of patients. There has never been any evidence that intentionally delaying tracheostomy improves outcomes; on the contrary, pooled evidence indicates that delayed tracheostomy is associated with higher incidence of long-term airway stenosis.15,22,23 We look forward to following this cohort of patients to assess additional and longer-term outcomes in the months ahead.
Analysis of early vs late tracheostomy is nuanced, both academically and clinically, by one of the central questions that determines timing of tracheostomy: how long do we estimate this patient will require ventilator support?13,14 One of the study’s principal limitations is that the answer to this is an educated guess, even in the minds of the most experienced clinicians, especially those in the heat of a pandemic of a virus whose natural history was vastly unknown in the early days. We collaborated with our colleagues in interventional pulmonology and critical care to innovate a protocol for earlier tracheostomy; the inherent bias in the study population is that the patients in this cohort were intentionally selected by a standardized set of criteria. Additionally, the study was limited intentionally by its defined time points at which the analysis was performed, namely 2 months in the earliest days of this pandemic. Thus, some patients remained mechanically ventilated at the end of the study period, and longer-term outcomes could not be assessed.
As the onslaught from the first surge of the pandemic settles, this analysis of data from our experience in the early months of the pandemic provides an opportunity to reconsider guidelines for tracheostomy for patients with COVID-19. These data demonstrate that with thoughtful selection of patients, there is no countervailing evidence to recommend categorically delaying tracheostomy in this patient population. When aligned with emerging evidence about the timeline of infectivity of SARS-CoV-2, thoughtful selection of patients can optimize outcomes from tracheostomy while keeping safe the clinicians who are committed to their care.
Accepted for Publication: October 26, 2020.
Published Online: December 17, 2020. doi:10.1001/jamaoto.2020.4837
Corresponding Author: Paul E. Kwak, MD, MM, MSc, NYU Voice Center, 345 E 37th St, Ste 306, New York, NY 10016 (firstname.lastname@example.org).
Author Contributions: Dr Kwak and Mr Connors had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Kwak, Timen, Sureau, Persky, Angel, Amin.
Acquisition, analysis, or interpretation of data: Kwak, Connors, Benedict, Timen, Wang, Zhang, Youlios, Persky, Rafeq, Amin.
Drafting of the manuscript: Kwak, Connors, Timen, Youlios.
Critical revision of the manuscript for important intellectual content: Kwak, Benedict, Timen, Wang, Zhang, Youlios, Sureau, Persky, Rafeq, Angel, Amin.
Statistical analysis: Connors, Benedict, Timen, Wang, Zhang, Persky.
Administrative, technical, or material support: Kwak, Timen, Youlios, Sureau, Persky, Angel, Amin.
Supervision: Kwak, Persky, Amin.
Other—clinical support: Rafeq.
Conflict of Interest Disclosures: None reported.