Effect of timing of intubation on clinical outcomes of critically ill patients with COVID-19: a systematic review and meta-analysis of non-randomized cohort studies

Back to the poliomyelitis epidemic of 1952, Dr. Henry Lassen from Copenhagen had reportedly noted that his American colleagues used to “put their patients in the respirators far too early” [1]. Therefore, since the birth of critical care, the optimal timing of intubation of patients with severe acute hypoxemic respiratory failure seems to be a matter of debate. This debate is reinforced during the ongoing epidemic of coronavirus disease 2019 (COVID-19).

Since the early phase of the COVID-19 epidemic, several guidelines from China [2], United Kingdom [3], United States of America [4] and Australia [5] recommend early intubation of critically ill patients with COVID-19 as a means to protect health care workers from cross-infection and to avoid complications (including cardiac arrest) associated with “crash” intubations. Experts of clinical respiratory physiology seemed to back this approach with notions that early intubation might prevent ensuing patient self-inflicted lung injury [6]. However, on the basis of physiological principles [7], other experts argued against early intubation [8]. Therefore, there seemed to be reasonable arguments in favor of either an early or a late intubation approach in COVID-19 and relevant studies were subsequently planned to address this clinical question.

Several studies reporting on outcomes of patients with COVID-19 undergoing early versus late intubation have indeed been published. However, the accumulated evidence has not yet been synthesized. Thus, we carried out a systematic review and meta-analysis in an attempt to investigate the effect (if any) of timing of intubation on clinical outcomes of critically ill patients with COVID-19.

We reported the current systematic review and meta-analysis in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [9]. We prespecified inclusion criteria, methods of data synthesis and outcomes in a protocol registered with PROSPERO (CRD42020222147) and available online [10].

Figure 1 shows the flow diagram for study selection. A total of 12 studies [13, 14, 18‐27] from Africa, Asia, Europe and America, involving 8944 critically ill patients (7639 early, 1305 late) with COVID-19, were incorporated in our meta-analysis. Table 1 and Additional file 1: Supplementary Table 2 summarize the characteristics and risk of bias assessment of the included studies, respectively.

By incorporating data from 12 studies involving almost 9000 critically ill patients across almost all continents, our systematic review and meta-analysis showed that timing of intubation may have no effect on all-cause mortality, duration of MV, ICU length of stay and renal replacement therapy.

Our systematic review identified variability in the literature regarding the definition of early intubation among the included studies, as presented in Table 1. In an attempt to increase homogeneity, we (inevitably arbitrarily) considered as “early” the intubation occurring within 24 h from ICU admission. We used a specific time threshold from ICU admission as a criterion for defining early/late intubation because this approach has been previously used in the literature both before [28] and during [25] the COVID-19 period. Similarly and interestingly, the Intensive Care National Audit and Research Centre (ICNARC) reports from the United Kingdom provide specific data on patients undergoing intubation within 24 h from ICU admission [29]. Also, given that SARS-Cov-2 virus predominantly affects the respiratory system, one could presume that ICU admission might serve as a surrogate of increased oxygen requirements and subsequent consideration of intubation. Indeed, acute hypoxemic respiratory failure was the main criterion for ICU admission in the largest study included in the meta-analysis [20]. That being said, we acknowledge that there might be variability among centers regarding criteria for ICU admission stemming, for example, from availability of resources, such as beds. In an attempt to address this concern, we used an alternate definition of early/late intubation having a prior trial of HFNC/NIV as criterion.

Regardless of the definition of early/late intubation (i.e., based on a specific time threshold from ICU admission or of a prior trial of HFNC/NIV) used in this meta-analysis, we found no statistically detectable difference on all-cause mortality between patients with severe COVID-19 undergoing early versus late intubation. Also, all-cause mortality was higher in the early than the late or no intubation group. These findings may not support the recommendation in favor of early over late intubation made by several international guidelines [2‐5]. Despite the above guidelines, clinicians caring of patients with COVID-19 seem to become eager to favor a wait-and-see strategy over time. Indeed, in a multicentre study from three European countries, involving 4244 critically ill patients with COVID-19, the percentage of patients receiving invasive MV descended from 82 to 68% [16]. A similar trend was reported in a large study from the USA [27]. This shift in intubation strategy from early to late, driven by clinical gestalt, seems to be justified by the findings of this meta-analysis.

We found no statistically detectable difference between patients undergoing early versus late intubation in terms of morbidity, namely duration of MV, ICU length of stay and renal replacement therapy. These findings referring to severe respiratory failure associated with COVID-19 are not in line with findings from observational studies on acute respiratory distress syndrome (ARDS) not associated with COVID-19. The latter studies reported that delaying intubation of critically ill patients with ARDS may be associated with adverse outcomes [28, 30, 31]. This is an interesting observation, which fuels the skepticism regarding the potential differences between ARDS associated with versus without COVID-19 [32, 33].

Our work has limitations. Firstly, as concerns to the outcomes of duration of MV and ICU length of stay, we noted considerable statistical heterogeneity. This probably reflects clinical heterogeneity due to variability among included studies regarding patient population characteristics and clinical practices. Secondly, we did not perform a trial sequential analysis. Thirdly, our meta-analysis is based on data from observational studies, which may suffer from residual confounding. Especially, we could not preclude that confounding by indication may be at play; i.e., sicker patients (with worse prognosis) might get intubated earlier than those with less severe disease. However, we attempted to assess risk of bias of included observational studies and carried out a subgroup analysis of studies with low risk of bias. Fourthly, our subgroup analysis of patients undergoing early versus late or no intubation may be limited by the bias that patients who were not eventually intubated might be, at the first place, less ill than those who were eventually intubated. Finally, the “late intubation” group might include both patients undergoing a time-limited trial of HFNC/NIV and patients remaining on HFNC/NIV for as long as possible. This should be taken into consideration when interpreting the results presented in Fig. 4.

The synthesized evidence of almost 9000 patients suggests that timing of intubation may have no effect on mortality and morbidity of critically ill patients with COVID-19. These results might justify a wait-and-see approach, which may lead to fewer intubations. Relevant guidelines may therefore need to be updated.