Tracheostomy for COVID-19 Respiratory Failure

Multidisciplinary, Multicenter Data on Timing, Technique, and Outcomes

Kamran Mahmood, MD, MPH; George Z. Cheng, MD, PhD; Keriann Van Nostrand, MD; Samira Shojaee, MD, MPH; Max T. Wayne, MD; Matthew Abbott, MD; Darrell Nettlow, MD; Alice Parish, MSPH; Cynthia L. Green, PhD; Javeryah Safi, MD; Michael J. Brenner, MD; Jose De Cardenas, MD


Annals of Surgery. 2021;274(2):234-239. 

In This Article


All adult patients with COVID-19 respiratory failure who required tracheostomy between February 1, 2020 and September 4, 2020 across 5 systems in United States were included. Data on outcomes were collected and reviewed last on January 15, 2021. Tracheostomies were performed by several specialties: interventional pulmonology, general surgery, thoracic surgery, otorhinolaryngology, and neurocritical care intensivists. Participating institutions included Duke University (Durham, NC), University of Michigan (Ann Arbor, MI), University of California at San Diego (San Diego, CA), Emory University (Atlanta, GA), and Virginia Commonwealth University (VCU; Richmond, VA). Institutional review board (IRB) approval was obtained from every institution (Duke IRB protocol # 00106067). Data were extracted from review of electronic medical records and included patient and disease characteristics; intra and post-procedural data; hospital course including ventilator weaning and length of stay; and outcomes including adverse events and survival. All data were entered into a REDCap database maintained by Duke University for secure web-based data capture from different institutions.

Statistical Analysis

Continuous data are presented as the median with 25th and 75th percentiles (Q1-Q3), whereas categorical data are displayed as counts with percentages. Comparisons between continuous variables were made using the nonparametric Wilcoxon rank sum test, and the Kruskal-Wallis test was used if >2 groups were compared, as none of the variables met normality assumptions for parametric tests. The Wilcoxon signed rank test was used to compare matched pairs. Categorical variables were assessed using the χ 2 test. SAS version 9.4 (SAS Institute, Inc., Cary, NC) was used for all analyses, and a two-sided P value <0.05 was considered statistically significant.