The effect of tracheostomy on extracorporeal membrane oxygenation outcomes

The optimal timing for pursuing tracheostomy in patients with prolonged mechanical ventilation with either veno-arterial (VA) or veno-venous (VV) extracorporeal membrane oxygenation (ECMO) is a discussion of risk versus benefit. Depending on the etiology, cardiothoracic surgical patients carry some of the highest risk for respiratory failure postprocedure. Given that patients with end-stage cardiopulmonary status may be fraught with substantial comorbidities, it is critically important to manage the risk-benefit profile of performing a tracheostomy procedure on a patient requiring ECMO support. These cohorts have risk factors that may depend on each patient's inflammatory state, lung de-recruitment peri-procedure and postprocedure and bleeding requiring transfusions to name a few. We provide a descriptive analysis of ECMO patients on both VA and VV configurations who survived to hospital discharge receiving tracheostomy either during or after their ECMO course.

A retrospective single-institutional study collected all consecutive patients age 18 and above who received any form of ECMO between 2016 and 2020. Five hundred forty-five patients were screened based on having received ECMO. Patients with mixed EMCO modality were excluded due to heterogeneity of disease process. A total of 521 patients received either VV or VA ECMO. A total of 54 patients received tracheostomy and had sufficiently clean data for analysis. Tracheostomy patients were compared based on survival to discharge, tracheostomy surgical complications, ECMO duration, ECMO configuration, inotrope and vasopressor use, transfusion rates, total ventilator days, total days on intravenous sedation, and history of cardiotomy or heart transplant were assessed. Baseline characteristics of race, age, gender, and body mass index (BMI) were also collected.

A total of 54 patients received tracheostomy. Twenty-nine of those patients received tracheostomy during the course of their ECMO, of whom 13 were on VV ECMO, 16 on VA ECMO. Another 25 patients underwent tracheostomy after successful ECMO explant; 8 of those were VV ECMO with the remaining 17 were on VA ECMO before explantation, with mean delay to tracheostomy, 10 and 19 days after explant between both modalities, respectively. A statistically significantly greater proportion of VV ECMO patients received a tracheostomy at any point versus VA ECMO patients (25.93% vs. 8.35%, p ≤ .0001). No statistically significant difference was noted in timing of tracheostomy when stratified by EMCO modality (VA 51.51% after explant vs. VV 38.10% after explant, p = .33). There was a greater frequency of minor tracheostomy complications in patients who were on ECMO at the time of their tracheostomy (p = .014) than in those who received their tracheostomy after being explanted. However, these minor complications did not contribute to a change in survival to hospital discharge (p = .58). Similarly, the small number of major complications (n = 13) did not impair survival to hospital discharge (p = .84). Finally, mean duration of ECMO was longer in those who received tracheostomy during ECMO versus after ECMO. (488.45 vs. 259.72 h, p < .01).

Tracheostomy is known to increase patient mobility, clinical participation, and overall decrease in sedation use. Pursuing tracheostomy during ECMO is feasible, does not result in major bleeding, and is associated with only minor complications that overall do not decrease survival. While there is an increased duration of ECMO support in the tracheostomy cohort, this may be due to existing patient conditions, and may not be causal. Research is needed to further determine the external patient factors and specific timing to optimize both VV and VA ECMO courses.

We hope that our analysis will pave the initial pathway for an evidence-based guideline on optimal timing of tracheostomy in ECMO patients, whether initiated during or after ECMO and taking into consideration ECMO configuration, its expected duration, and patient comorbidities.