Outcomes of Combined Percutaneous Tracheostomy and Endoscopic Gastrostomy Tube Placement in COVID-19 Patients

Background: A signicant number of patients with severe respiratory failure related to COVID-19 require prolonged mechanical ventilation. Minimal data exists regarding the timing, safety, and ecacy of combined bedside percutaneous tracheostomy and endoscopy gastrostomy tube placement in these patients. The safety for healthcare providers is also in question. This study’s objective was to evaluate the effectiveness and safety of combined bedside tracheostomy and gastrostomy tube placement in COVID-19 patients. Methods: This is a single arm, prospective cohort study in patients with COVID-19 and acute respiratory failure requiring prolonged mechanical ventilation who underwent bedside tracheostomy and percutaneous endoscopic gastrostomy tube placement. Detailed clinical and procedural data were collected. Descriptive statistics were employed and time to event curves were estimated and plotted using the Kaplan Meier method for clinically relevant pre-specied endpoints. Results: In this study, we prospectively evaluated ICU patients diagnosed with SARS-CoV-2 undergoing percutaneous tracheostomy and gastrostomy tube placements by the Interventional Pulmonology (IP) service at the Massachusetts General Hospital in order to assess the impact of these procedures on patients with COVID-19. Included in this analysis are comprehensive details of patient demographics, comorbid conditions, and ICU-specic data, such as laboratory values, ventilator settings, and sedative requirements. We provide in-depth procedural data particularly pertinent in the COVID-19 pandemic, including number of personnel required per case, apnea time for tracheostomy placement, and both mouth-to-stomach and total procedural time for gastrostomy tube placement, which to our knowledge have not yet been reported in COVID-19 patients. We include 14, 30, and 60-day mortality rates, ICU and hospital lengths of stay, and ultimate disposition in over 98% of patients. Finally, we conducted time-to-event analyses for clinically meaningful prespecied endpoints including time to weaning from mechanical ventilation, tracheostomy downsizing, decannulation, and hospital discharge.


Background
The Coronavirus disease 2019  pandemic caused by the SARS-CoV-2 virus has taken the world by storm, infecting over 52.3 million patients globally as of November 12, 2020. In the United States (US) alone, over 10.4 million cases have been reported (1). This highly contagious respiratory virus typically causes mild symptoms such as fever, cough, fatigue, and dyspnea, however a subset of patients will develop critical illness, often including acute respiratory failure, acute respiratory distress syndrome (ARDS), and multiorgan failure (2). In a retrospective review of 191 patients hospitalized with COVID-19 in Wuhan, China, 32 (16.7%) required invasive mechanical ventilation with a mean initiation at 14.5 days from symptom onset (3). A meta-analysis of 3,062 patients with COVID-19 across 38 studies showed an incidence of ARDS in 19.5% (4). Of 73 ventilated patients with COVID-19 and ARDS in Milan, Italy, 45.2% continued to need invasive mechanical ventilation at day 19 of their intensive care unit (ICU) course (5).
In 21 critically ill COVID-19 patients in Washington, US, 71% patients required mechanical ventilation and 38% of those were unable to be weaned by day 25 (6).
These data suggest that a signi cant number of patients admitted to the ICU with severe respiratory failure will require prolonged mechanical ventilation. The pressing question is how to provide this support in the most e cient, safe and effective fashion while minimizing the risk of transmission to healthcare personnel (HCP).
Tracheostomy placement is used in patients with prolonged respiratory failure for several reasons -it improves patient communication, work of breathing, need for sedation, and overall patient comfort (7,8).
They can be placed either percutaneously or surgically, with most centers deferring to their available human capital and subsequently preferred technique. The percutaneous approach has been shown to require less procedural time, has less wound infection and scarring, and is less expensive than the surgical technique. No difference in mortality or intra-procedural complication rates have been shown (9,10). The percutaneous technique is typically performed at bedside in the ICU while surgical placement is usually done in the operating room (OR). Gastrostomy tubes are often required in these patients as well to provide nutritional support, particularly in patients who are not able to be weaned from mechanical ventilation (11). If longer term support is needed, gastrostomy tubes are frequently preferred over nasogastric or orogastric tubes for stability as well as to mitigate the risk of sinusitis, pharyngeal discomfort, epistaxis, and short-term aspiration (12,13). Gastrostomy tubes are placed by a wide array of physicians, including gastroenterologists, general surgeons, interventional pulmonologists, and interventional radiologists, and utilize various techniques, such as endoscopic, radiologic, or surgical placement (14). Endoscopic placement can be done at bedside in the ICU while surgical or radiologic placement is typically performed in the OR or procedural suite.
A signi cant portion of patients with COVID-19 associated respiratory failure who have survived the acute phase of disease are now requiring prolonged care. The exact timing and technique for tracheostomy placement is unclear, with minimal evidence-based guidelines available. Several societal guidelines and expert opinion papers recommend tracheostomy placement after day 10-14 of mechanical ventilation (15)(16)(17). The choice of procedure type and location are generally left to the discretion of the provider with priorities placed on utilizing the most experienced operator while balancing the risks to both provider and patient. Similar questions exist regarding optimal timing and technique for gastrostomy tube placement. As these procedures are high-risk for aerosol generation, the goal should be to perform them in a safe and effective manner while minimizing risk to HCP involved. This can involve steps such as minimizing the number of personnel in the room at the time of placement, utilizing experts in each procedure in order to decrease overall procedural length, performing these procedures at bedside in the ICU to decrease the number of patient transfers, and performing both procedures during one anesthetic episode.
In this study, we prospectively evaluated ICU patients diagnosed with SARS-CoV-2 undergoing percutaneous tracheostomy and gastrostomy tube placements by the Interventional Pulmonology (IP) service at the Massachusetts General Hospital in order to assess the impact of these procedures on patients with COVID-19. Included in this analysis are comprehensive details of patient demographics, comorbid conditions, and ICU-speci c data, such as laboratory values, ventilator settings, and sedative requirements. We provide in-depth procedural data particularly pertinent in the COVID-19 pandemic, including number of personnel required per case, apnea time for tracheostomy placement, and both mouth-to-stomach and total procedural time for gastrostomy tube placement, which to our knowledge have not yet been reported in COVID-19 patients. We include 14, 30, and 60-day mortality rates, ICU and hospital lengths of stay, and ultimate disposition in over 98% of patients. Finally, we conducted time-toevent analyses for clinically meaningful prespeci ed endpoints including time to weaning from mechanical ventilation, tracheostomy downsizing, decannulation, and hospital discharge.

Study Design and Participants
This was a prospective, single-arm, cohort study conducted in the ICU at the Massachusetts General Hospital. Institutional Review Board approval was obtained (#2020P001427). Adult ICU patients who underwent percutaneous tracheostomy with or without percutaneous endoscopic gastrostomy (PEG) for acute respiratory failure due to SARS-CoV-2 were included. Patients who received surgical tracheostomy placement or surgical or radiologic gastrostomy tube placement were excluded.

Data Collection
Demographic, clinical, procedural, laboratory, radiologic, and outcome data were obtained from the electronic medical record system using a standardized collection process. All data were reviewed by two separate physicians (EF and CO).

Procedural Technique
Tracheostomies were performed via the percutaneous dilational bronchoscopic-guided technique utilizing a Ciaglia Blue Rhino® kit (Cook Critical Care, Bloomington, IN, USA) (18). The patient was sedated and received neuromuscular blockade, positioned, and prepped in sterile fashion. All equipment was opened and prepared. A high e ciency particulate air (HEPA) lter was already in place between the ventilator and tubing. The ventilator was set to volume control at 100% fraction of inspired oxygen (FiO 2 ). All personnel other than the primary operators then exited the room and were immediately available outside the glass doors. After careful landmark determination, local anesthetic was injected, the incision carried out, and the soft tissue bluntly dissected. At this point, in a highly coordinated fashion, the ventilator was paused, and the proximal limb disconnected at the machine. This was done intentionally to reduce Results Fifty-eight patients were included from April 5, 2020, through June 15, 2020 ( Fig. 1). Follow up was extended through August 6, 2020. Demographics and clinical characteristics are shown in Table 1. The median age of patients was 59 years old with a body mass index (BMI) of 27.6. Nearly 90% of the patients received pronation therapy and 5% were either receiving or had received extracorporeal membranous oxygenation (ECMO). An additional 46% of patients underwent ECMO consultation but were found to have an absolute or relative contraindication such as BMI, advanced age or low likelihood of recovery. All patients were mechanically ventilated with an average positive end-expiratory pressure (PEEP) of 8.0 cm H 2 O and FiO 2 of 35% the day prior to the procedure. Fifty-six patients were persistently SARS-CoV-2 positive at the time of the procedure.
aerosolization of viral particles during high risk portions of the procedure. Voluntary apnea was con rmed, and a timer started. The bronchoscope was inserted through the endotracheal tube (ETT) and after rapid ETT cuff de ation, the tube was quickly pulled back to the subglottic space. In a few instances, rapid desaturation was seen. When the oxygen saturation reached 80%, the procedure was interrupted, the ETT was advanced, the circuit closed, and ventilation was brie y resumed. Once preprocedural oxygen saturation was achieved, the circuit was disconnected as described above and the procedure was resumed and nished. Rapid and coordinated Seldinger-technique placement of the tracheostomy tube was carried out, and after cuff in ation, the tracheostomy tube was reconnected to the ventilator circuit. After con rming adequate ventilation and seal of the respiratory system, the team moved to the abdomen.
Gastrostomy tubes were placed via the endoscope-bronchoscope-guided percutaneous method using the 24-French Boston Scienti c Pull PEG kit (Boston Scienti c Corporation, Natick, Massachussetts) (14). Adequate transillumination and digital pressure in the abdomen served as con rmatory measures. The endoscopic equipment used then underwent high-level disinfection as recommended by the manufacturer after each use. These protocols have been proven to have excellent antimicrobial activity (19,20).

Statistical Analysis
Standard descriptive statistics were employed using median and interquartile range (IQR) for continuous variables and frequency and percentage for categorical variables. Time to event analyses for prespeci ed endpoints -weaning from mechanical ventilation, tracheostomy tube downsizing, decannulation, and hospital discharge -were estimated and plotted using the Kaplan Meier method. We also present the estimated median and time to event as well as 95% con dence interval for these outcomes. Statistical analyses were run using IBM SPSS V25 (Arnonk, NY). Data are median (IQR) or n (%). BMI = body mass index, COVID-19 = coronavirus disease 2019, SOFA = sequential organ failure assessment, ECMO = extracorporeal membranous oxygenation a Antiplatelet agents include aspirin (325 mg), clopidogrel, prasugrel, ticagrelor b Therapeutic anticoagulation de ned as treatment dose heparin, enoxaparin, warfarin, apixaban, rivaroxaban c Prophylactic anticoagulation de ned as prophylactic dose heparin Eighty-six percent of patients received both tracheostomy and PEG tube placements with 13.8% receiving tracheostomy alone. Of the 50 patients who had both tracheostomy and PEG tubes placed, 49 were completed in the same procedure during the same anesthetic episode. All procedures were done at bedside with no patient transfer required out of the ICU. A median of 3.0 HCP total were present in the room per procedure. Nursing and respiratory therapists were immediately available outside of the room to provide assistance if needed. Median apnea time during tracheostomy placement was 90 seconds (IQR 55.8-120). Mouth-to-stomach time for PEG placement was 30 seconds (IQR 25-53.5) and total PEG procedure time was 14.5 minutes (IQR 10.6-17). Procedural complications were as follows: minor bleeding in 12 patients, site irritation in 6, and repeat procedure needed in 5. There were no serious complications such as perforation of adjacent structures or periprocedural death in any patient. Procedural data is summarized in Table 2. Data are median (IQR) or n (%). ICU = intensive care unit, PRBC = packed red blood cell, PEG = percutaneous endoscopic gastrostomy Median ICU length of stay (LOS) was 29 days with a median LOS of 10 days post-procedure (see Table 3). Total hospital LOS was 45 days. Nearly 88% of patients were weaned from mechanical ventilation post-procedure at a median of 9 days. Thirty-nine patients were downsized and 49 were ultimately decannulated. Time to downsizing and decannulation was 17 days and 25 days, respectively (see Fig. 2). Mortality rates at 14, 30, and 60 days were 5.2% 6.9%, and 10.3%, respectively. Nearly 90% of patients were discharged alive from the hospital.

Discussion
The COVID-19 pandemic has profoundly impacted patient care worldwide as hospitals and HCP innovate and adapt to the new realities with which they are faced. Critically ill patients with COVID-19 will often need prolonged mechanical ventilation as they recover, raising the question of how this should best be achieved to maximize patient bene t while minimizing risk to HCP. Most societies agree that tracheostomy placement should occur between days 10 and 14 of mechanical ventilation, however there is no consensus regarding details such as optimal technique or procedural location. Additionally, minimal data has been published regarding COVID-19 patients and tracheostomy and gastrostomy tube placements. To our knowledge, this is the only study to report clinical outcomes as well as in-depth demographic and periprocedural data in critically ill COVID-19 patients who underwent combined tracheostomy and gastrostomy tube placements.
This study highlights several important points speci cally pertaining to tracheostomy and gastrostomy tube placements in COVID-19 patients. First, no transfer out of the patient's room was required for any procedure. This has a profound impact currently as it eliminates exposure to multiple HCP -physicians, nursing staff, respiratory therapists, transport staff, and operating room or procedure suite staff. It also decreases contamination and need for terminal cleaning of multiple healthcare areas.
Second, the majority of procedures were performed in the same anesthetic episode, which has multiple bene ts. Not only does it decrease the number of times a patient is sedated, it also reduces the exposure to any involved HCP. In this study, these procedures were done with only 2-3 HCP per procedure (including combined procedures), without any incidents. This was designed intentionally to expose the least number of HCP to SARS-CoV-2. In follow up to this study, the physicians primarily involved in performing these procedures tested negative for SARS-CoV-2 IgM and IgG antibodies, indicating they had likely not been exposed despite performing the majority of procedures in currently positive patients.
Third, our ability to provide nutritional support to patients with critical illness as the disease becomes chronically debilitating is likely to impact their overall outcomes. The published data on gastrostomy tubes and nutrition during the COVID-19 pandemic is very limited so far. A retrospective review involving six centers in New York described all cases of endoscopy performed during the COVID-19 pandemic. This included 605 endoscopies in 545 patients with and without COVID-19. Of the 84 endoscopies done on COVID-19 positive patients, nearly one-third of these were performed for PEG or nasogastric tube placement (21). This critically ill patient population is likely to be in a hypercatabolic state with increased energy expenditure linked to ventilatory work during their ICU stay. Furthermore, at the time of admission to the hospital, their nutritional status may already be compromised by preexisting factors such as anorexia, dyspnea, anosmia, dysgeusia, con nement, and limited access to food. During their prolonged hospitalization, hypermetabolism and physical immobilization are likely to worsen. The European Society for Clinical Nutrition and Metabolism (ESPEN) makes recommendations that favor enteral nutrition over parenteral nutrition, including in the prone position. Nutrition support should be continued after extubation until the patient resumes su cient oral intake (22,23).
Fourth, tracheostomy placement in these patients potentially allows for more rapid weaning of sedating medications, which impacts incidence of delirium, weaning from mechanical ventilation, ICU LOS, and development of critical illness polyneuropathy. Multiple trials have demonstrated the importance of daily interruption of sedatives as well as spontaneous awakening and spontaneous breathing trials in mechanically ventilated ICU patients (24,25). Tracheostomy has been shown previously to decrease the need for intravenous sedatives and to facilitate patient autonomy earlier compared to continued ventilation via an endotracheal tube (26). Our data suggests this may be similar in critically ill COVID-19 patients.
Finally, 19% of patients required pre-procedural transfusions and over 40% were on therapeutic anticoagulation that had to be brie y discontinued prior to the procedure(s). In performing combined tracheostomy and gastrostomy tube placements, both the number of transfusions and times anticoagulation was interrupted were potentially decreased by half.
In this study, the 60-day mortality rate was 10.3% (n = 6) (Fig. 1). This nding is lower than that initially reported in other series of critically ill, mechanically ventilated patients, which was anywhere between 50-90%, but is consistent with the trend of decreasing mortality in subsequent reports (27,28). While the median Sequential Organ Failure Assessment (SOFA) scores in our cohort was only 5, we feel this is not a true re ection of the overall severity of illness. These scores were measured immediately prior to the procedure and are not necessarily the highest overall score for the ICU stay. Additionally, mental status may not have been accurately assessed in the majority of patients due to sedation. This has been described previously as the CNS component of the SOFA score is the least accurate (29, 30). In a cohort of 59 critically ill COVID-19 patients on mechanical ventilation or mechanical ventilation and ECMO, the average SOFA score was only 6.5 (31). Nearly 52% of our patients were considered for ECMO and 89% had received pronation therapy for severe respiratory failure, which we feel is a more accurate re ection of the overall severity of illness of the cohort.
Over 98% of patients alive at 60 days had been weaned from mechanical ventilation and 94% of this group had been decannulated. In a cohort of 1890 COVID-19 patients with tracheostomies, 23.7% died by 30-day follow up. Only 52% of patients were weaned from mechanical ventilation; 81% of those patients were decannulated (27). In our cohort, median time to weaning was 9 days, to downsizing was 17 days, and to decannulation was 25 days. Median time to hospital discharge was 25 days. Time to event curves are shown in Fig. 2. Prior to tracheostomy and gastrostomy tube placements, patients had a median ICU LOS of 19 days. Post-procedure, median ICU LOS was 10 days. This data supports the use of tracheostomy tubes in COVID-19 patients who require prolonged mechanical ventilation to facilitate weaning and transition out of the ICU to a rehabilitation environment for further recovery. In our cohort, over 80% of our patients required further care at a skilled nursing or long-term acute care facility with excellent overall results.
While bleeding did occur in 20% of patients, this was classi ed as minor and did not require surgery or embolization. It is important to recognize the coexisting factors in these patients including the frequency of acute renal failure and use of anticoagulation. Fortunately, there were no major complications, such as organ perforation or death, in any patient.

Conclusions
The COVID-19 pandemic has signi cantly impacted the delivery of care in critically ill patients. A signi cant number of these patients will require prolonged mechanical ventilation as they recover from their illnesses. Tracheostomy and gastrostomy tubes are important facilitators of weaning from ventilation and transition out of the ICU into a rehabilitation environment. Here, we report in-depth demographic, clinical, procedural, and outcome data on a single-arm cohort of critically ill patients with COVID-19 who underwent tracheostomy and gastrostomy tube placements. Importantly, nearly 90% of these patients were discharged alive from the hospital and 94% of patients weaned from mechanical ventilation were decannulated. This study supports the use of tracheostomy and gastrostomy tube placement in critically ill COVID-19 patients. Clinical outcomes in 58 COVID-19 patients with tracheostomy and PEG tube placements apatient had preexisting quadriplegia Time to event curves