Time to Extubation in COVID-19 and Non–COVID-19 ARDS

In late 2019 and early 2020, the world was watching and monitoring a new virus that reportedly started in Wuhan, China.¹ The virus (COVID-19) was easily transmittable between humans through respiratory droplets. Scientists worked diligently to develop new antiviral strategies as the virus spread exponentially within a population and between populations around the world. At the same time, health care practitioners found themselves challenged to find the optimum therapies.^2–4 A common development in hospitalized patients infected with COVID-19 during the early phase of the pandemic was the progression of respiratory distress leading to failure. Since the late 1990s, there has been an internationally accepted constellation of symptoms associated with the diagnosis of ARDS. In addition, many clinicians adopted the management protocols defined in these early landmark studies. As COVID-19–induced ARDS began to present itself in hospitals across the globe, clinical researchers began to question if COVID-19 ARDS was similar to other forms of ARDS.⁵

In this issue of Respiratory Care, Gregory Burns and colleagues⁶ examine one outcome of interest, specifically time to extubation. Secondary outcomes of interest, such as time to weaning readiness and rate of P_aO2/F_IO2 improvement, were also examined in the reported cohorts. The study subjects were retrospectively identified from examination of the electronic medical record at a single, academic medical institution. A total of 73 subjects, admitted between March 1–August 12, 2020, were identified based upon the inclusion criteria established a priori by the research team. Thirty-seven subjects were admitted for ARDS associated with COVID-19, while 36 subjects had ARDS from other sources. All subjects were managed with established ARDS Network ventilator protocol and required > 48 h of mechanical ventilation. Patients < 18 y of age, those who received a tracheostomy, or those who were inter-facility transfers were excluded.

The research team found that mechanical ventilation duration was significantly greater in COVID-19 ARDS than the comparative cohort of ARDS from other sources. In addition, the rate of P_aO2/F_IO2 improvement was significantly slower in the COVID-19 ARDS group. Looking further, the research team demonstrated a significantly longer time associated with reaching readiness to wean status in the COVID-19 ARDS cohort. In a different study, Bain and colleagues⁵ reported on similarities and differences between COVID-19–induced ARDS and other forms of ARDS (viral, bacteria, non–culture positive pneumonia) during a similar period. Bain analyzed the medical records of 27 subjects with COVID-19–induced ARDS compared to viral-induced ARDS (n = 14), bacterial-induced ARDS (n = 21), and culture-negative pneumonia-induced ARDS (n = 30). In their study, the investigators found that subjects with COVID-19 were associated with a higher body mass index (obesity), and were more likely to be Black or residents of skilled nursing facilities.⁵ They found no significant difference in static compliance, hypoxemic indices, or carbon dioxide clearance between COVID-19 ARDS and other forms of ARDS. However, subjects with COVID-19 had lower minute ventilation and longer duration of ventilation with similar 60-d mortality. Bain et al concluded that the differences observed in their study cohorts did not merit different management strategies for COVID-19–induced ARDS. Similar findings as those reported by Burns and colleagues⁶ were also reported by other investigators who examined subjects during this early time period of the pandemic.^7,8 There were no reported randomized controlled trials during this period.

A limitation of this study is the small sample size. Although the study was stated to be adequately powered to observe a statistical difference in mechanical ventilation, the small sample size and an additional limitation of being from a single medical center make generalizing the findings to outside larger cohorts limited at best. Numerous studies reporting on outcomes from this time have shown substantial variability across centers and around the world. The variability in clinical practice and outcomes throughout the world during this early phase of the pandemic makes extrapolation of results or comparison of cohorts difficult.

Another limitation is the limited window established for this observational study. COVID-associated ARDS evolved significantly as the pandemic, our understanding of the virus, and development of new strategies of care and treatment have continued to favorably impact outcomes associated with this study population. As such, the relevance of these findings to current COVID-19 ARDS ventilated subjects requires further investigation.

A final limitation was the method selected for imputing missing values. The authors identified that some data used in the longitudinal analyses were imputed due to those parameters being missing in the electronic medical record. The authors employed a common technique of imputation involving last observation carried forward (LOCF). Although this technique can be found in numerous peer-reviewed articles, LOCF has been shown to produce potentially spurious analysis resulting from this methodology.⁹ The authors do not identify what data were missing and what percentage of the longitudinal data required LOCF. As such, we must be cautious in our acceptance of the findings.

The limitations notwithstanding, the authors should be acknowledged for their efforts in bringing forward this early snapshot of mechanical ventilation outcomes with comparison between COVID-19 ARDS and ARDS from other sources. A unique analytical approach was employed by the authors that examined the instantaneous probability of extubation in a longitudinal model. This is an intriguing aspect of their study that may warrant further exploration to determine the clinical or scientific utility of this metric. As reported, there appears to be only one other study that examined instantaneous probability of extubation in a similar population. That author reported no significant difference, contrary to the findings reported by Burns et al.⁶ Burns and colleagues correctly identified the challenges of comparing findings between studies when risk survival models are employed. Time-to-event analyses can be highly nuanced based upon the controlling variables employed in each model.

In conclusion, we have gained substantial knowledge regarding the COVID-19 virus since the early moments of this pandemic. Observational studies, such as this, are important to assist scientists and medical practitioners in hypothesis generation for future randomized controlled studies.