Background

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has become one of the largest known pandemic in human history affecting more than 233 million people across the globe [1]. Although the majority of individuals experience mild symptoms, approximately 5–15% develop respiratory failure and require invasive mechanical ventilation (IMV) [2,3,4,5,6]. Earlier reports of patients with coronavirus disease 2019 (COVID-19) on IMV described poor outcomes, with mortality rates as high as 45–74%, and 50% of patients requiring prolonged IMV (> 2 weeks) [4, 7,8,9,10,11].

A shorter ventilator time and ICU stay were particularly valuable during the COVID-19 pandemic, when intensive care units (ICUs) had insufficient ventilator and beds [12]. Tracheostomy was considered an attractive intervention to potentially reduce the time on a ventilator, length of ICU stay, and mortality [13, 14]. Nevertheless, most published guidelines in COVID-19 did not recommend performing early tracheostomy (ET) due to these early reports suggesting high mortality rates and high risk for possible virus transmission to health care workers during the tracheostomy procedure [15,16,17,18,19,20]. Unfortunately, most guidelines were published at the beginning of the pandemic without data to sustain them.

This year, several studies have attempted to investigate how ET affects COVID-19 outcomes [21,22,23]. However, whether ET improves COVID-19 outcomes is still controversial [24,25,26]. Thus, our objective was to systematically appraise the existing COVID-19 studies examining the impact of ET on the primary outcomes of duration of IMV, duration of ICU stay, and overall reported mortality and secondary outcomes of ventilator-associated pneumonia (VAP), time from tracheostomy to ventilator weaning, and duration of sedation.

Methods

We conducted a systematic review and meta-analysis according to the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement (see Additional file 1 for PRISMA checklist) [27]. The protocol for this review was registered on the International Platform of Registered Systematic Review and Meta-analysis Protocols database on August 23, 2021 (INPLASY202180088), and is available in full on inplasy.com (https://doi.org/10.37766/inplasy2021.8.0088).

Search strategy and study selection

Two investigators (YJ and BC) systematically searched Medline, Embase, Cochrane, and Scopus database from December 1, 2019, to August 24, 2021, which was the date of our last search. Search terms included (novel coronavirus OR SARS-CoV-2 OR COVID19 OR COVID-19) AND (tracheostomy OR tracheotomy) (see Additional file 2 for search strategy). We screened the reference lists of included articles. We also searched preprint servers (namely, medRxiv, bioRxiv, and Research Square) from December 1, 2019, to August 24, 2021.

There were no restrictions on language, location, or sample size for included studies. Two investigators (YJ and BC) independently screened both titles and abstracts to determine suitability based on our primary outcomes. Relevant full-text articles were retrieved and analyzed for eligibility. A third reviewer (XF) adjudicated discrepancies, when necessary.

Studies were included if they compared ET versus LT and provided data on at least 1 of our primary outcomes. Case reports, reviews, editorials, commentaries, and practice guidelines were excluded. Articles available only in abstract form or meeting reports were also excluded. The inclusion and exclusion criteria are given in detail in Additional file 3.

Data collection and quality assessment

Data collection was performed by two independent reviewers (YJ and BC) using a prespecified data extraction form. Disagreements were resolved by discussion and consensus. We collected the following data: first author and location, study period, publication format, type of study, timing of tracheostomy, number of patients, age, gender, the rate of percutaneous dilatation procedures, duration of IMV, duration of ICU stay, mortality, VAP, time from tracheostomy to ventilator weaning, duration of sedation, major complications related to tracheostomy, and transmission of SARS-CoV-2 from patients to health care workers.

The methodological quality of the selected articles was evaluated using the Newcastle–Ottawa quality assessment scale, whereby a higher score indicated higher methodological quality [28]. We assigned scores of 0–3, 4–6, and 7–9 for low, moderate, and high-quality articles, respectively.

Definitions and outcomes

We defined ET as a tracheostomy conducted 14 days or less after initiation of IMV. LT was any time thereafter. If a study defined ET after 14 days, we did not include the study in this review. In other words, ET/LT cut-off was defined as equal to or less than 14 days after initiation of IMV. We had 3 distinct primary outcomes: duration of IMV (from IMV initiation to discontinuation), duration of ICU stay (the number of days of stay in the ICU), and overall reported mortality (as reported at specific time points by study authors). Secondary outcomes included (1) VAP (according to study authors’ definitions of VAP), (2) time from tracheostomy to ventilator weaning (as defined by study authors), and (3) duration of sedation (the total number of days of sedation).

Statistical analysis

Meta-analysis was performed using Review Manager 5.4 (Revman, The Cochrane Collaboration, Oxford, UK). The estimation of combined continuous values and dichotomous values was expressed as mean differences (MD) or odds ratios (OR), respectively, with 95% confidence intervals (CI). When continuous values were presented as median and interquartile range (IQR), we calculated the mean and standard deviation (SD) as per Wan et al. [29]. We combined means and SDs from multiple groups into one group, when necessary, using the formula provided by StatsToDo (www.statstodo.com). A random-effects model was used to analyze data. Statistical heterogeneity was evaluated using the I2 statistic. I2 values > 0%, > 30%, > 50%, and > 75% were considered to indicate low, moderate, substantial, and considerable heterogeneity, respectively. If I2 was > 50%, we performed a sensitivity analysis by removing 1 study at a time (guided by the highest I2) until the sensitivity was below the threshold of 50% [30]. Additionally, we performed another sensitivity analysis by restricting the analysis to studies published in peer-reviewed journals. Since the included studies diverged by tracheostomy timing, we performed a subgroup analysis by dividing the studies into two groups according to the methodology of determining the timing of ET into studies that considered ET within the first 7 days of endotracheal intubation and studies that considered ET within 14 days of intubation. A p value < 0.05 was considered significant.

Results

Figure 1 shows the flow diagram of study selection process. A total of 14 studies [21,22,23, 31,32,33,34,35,36,37,38,39,40,41] (11 peer-reviewed and 3 preprints; 6 prospectively conducted and 8 retrospectively conducted; 8 single-center studies and 6 multicenter studies) from Asia, Europe and America, involving 2371 tracheostomized COVID-19 patients (938 in the ET group vs. 1433 in the LT group), were incorporated in our meta-analysis. The majority of the studies were conducted during the first wave of COVID-19 pandemic. The study authors defined ET as fewer than 7 days, up to fewer than 14 days post-IMV. Tracheostomy was performed via percutaneous or surgical techniques. The most frequent major complication was bleeding that required transfusion or surgical control. None of the health care workers tested positive or developed COVID-19 symptoms following tracheostomy in all 9 studies that provided this information. Table 1 summarizes the characteristics of the included studies. In the quality assessment of the 14 included studies, 13 were rated as high quality and one as moderate quality (Additional file 4: Table S2).

Fig. 1
figure 1

Enrollment flow diagram

Full size image
Table 1 Characteristics of studies included in systematic review
Full size table

Primary outcomes

Duration of IMV

Nine studies [21,22,23, 31, 32, 34, 35, 37, 40] provided data on duration of IMV. Substantial statistical heterogeneity was observed (I2 = 57%). ET was associated with decreased duration of IMV (2098 patients; MD − 9.08 days, 95% CI − 10.91 to − 7.26 days, p < 0.01) (Fig. 2).

Fig. 2
figure 2

Association of early tracheostomy with duration of invasive mechanical ventilation

Full size image

Duration of ICU stay

Seven studies [21,22,23, 32, 34, 35, 40] provided data on duration of ICU stay. Substantial statistical heterogeneity was detected (I2 = 67%). ET was associated with decreased duration of ICU stay (1224 patients; MD − 9.41 days, 95% CI − 12.36 to − 6.46 days, p < 0.01) (Fig. 3).

Fig. 3
figure 3

Association of early tracheostomy with duration of ICU stay

Full size image

Overall mortality

Thirteen studies [21,22,23, 31,32,33, 35,36,37,38,39,40,41] provided data on overall mortality. Mortality was reported at 30 days following tracheostomy [33], at 30 days following ICU admission [32, 35], at 60 days following intubation [39], at ICU [21, 40, 41] and hospital discharge [23, 31, 36, 38] and at an undefined time point [22, 37]. Moderate statistical heterogeneity was detected (I2 = 41%). There was no statistically detectable difference between patients undergoing ET versus LT regarding mortality (2343 patients; 32.1% vs. 29.3%; OR 1.09, 95% CI 0.79–1.51, p = 0.59) (Fig. 4).

Fig. 4
figure 4

Mortality outcome in early versus late tracheostomy

Full size image

Secondary outcomes

VAP

Two studies [21, 23] provided data on VAP. One study [21] defined VAP as pneumonia occurring > 48 h after endotracheal intubation and the other [23] defined VAP as clinical suspicion of ventilator-associated pneumonia with positive respiratory cultures that necessitated antibiotic administration. No statistical heterogeneity was detected (I2 = 0%). The incidence of VAP was lower in patients with IMV who underwent ET compared with LT (800 patients; 37.3% vs. 51.8%; OR 0.65, 95% CI 0.48–0.88, p < 0.01) (Fig. 5).

Fig. 5
figure 5

Ventilator-associated pneumonia in early versus late tracheostomy

Full size image

Time from tracheostomy to ventilator weaning

Six studies [21, 23, 32, 36,37,38] provided data on time from tracheostomy to ventilator weaning. Ventilator weaning was not defined in two studies [23, 37] and was defined in four studies as discontinuation of mechanical ventilation [21, 32, 36, 38]. Low statistical heterogeneity was detected (I2 = 30%). Patients undergoing ET had a numerically shorter time from tracheostomy to ventilator weaning, but this difference was not statistically significant (1715 patients; MD − 1.11 days, 95% CI − 2.63 to 0.41 days, p = 0.15) (Additional file 5: Figure S1).

Duration of sedation

Only one study [23] provided data on duration of sedation. As such, a pooled estimate could not be estimated. The test for heterogeneity was not applicable. There was a statistically significant difference between patients undergoing ET versus LT regarding duration of sedation (118 patients; MD − 7.60 days, 95% CI  − 10.57 to  − 4.63 days, p < 0.01). (Additional file 5: Figure S2).

Sensitivity analyses

We conducted sensitivity analyses on primary outcomes with substantial heterogeneity (I2 > 50%). For the duration of IMV, 1 study [32] had high heterogeneity and was removed for sensitivity analysis. ET was associated with decreased mechanical ventilation time (1998 patients; MD − 9.80 days, 95% CI − 11.39 to − 8.22 days, p < 0.01; I2 = 31%) (Additional file 5: Figure S3). For the duration of ICU stay, six studies [23, 31, 32, 34, 35, 40] remained after removing one with high heterogeneity. ET was associated with decreased ICU days (542 patients; MD − 7.57 days, 95% CI − 9.40 to − 5.74 days, p < 0.01; I2 = 0%) (Additional file 5: Figure S4).

The sensitivity analyses, restricted to studies published in peer-reviewed journals, found that ET was associated with decreased duration of IMV (1226 patients; MD − 9.54 days, 95% CI − 12.32 to − 6.76 days, p < 0.01; I2 = 65%) and duration of ICU stay (1048 patients; MD − 10.13 days, 95% CI − 14.27 to − 6.00 days, p < 0.01; I2 = 75%) (Additional file 5: Figures S5–S6). There was no statistically detectable difference on overall mortality between patients undergoing ET versus LT (1471 patients; 35.0% versus 29.8%; OR 1.24, 95% CI 0.81–1.89, p = 0.32; I2 = 49%) (Additional file 5: Figure S7).

Subgroup analyses

Studies were divided into two groups according to the methodology of determining the cut-off timing for ET, they were divided into studies that considered ET within the first 7 or 14 days of endotracheal intubation. ET was associated with shorter duration of IMV in studies defining ET as that done within 7 (799 patients; MD − 8.49 days, 95% CI − 10.94 to − 6.05 days, p < 0.01; I2 = 12%) or 14 (1981 patients; MD − 9.35 days, 95% CI − 11.36 to − 7.34 days, p < 0.01; I2 = 58%) days (Additional file 5: Figures S8–S9). ET was also associated with shorter duration of ICU stay in studies defining ET as that done within 7 (799 patients; MD − 8.40 days, 95% CI − 11.32 to − 5.48 days, p < 0.01; I2 = 0%) or 14 (1107 patients; MD − 9.75 days, 95% CI − 13.24 to − 6.27 days, p < 0.01; I2 = 71%) days (Additional file 5: Figures S10–S11). No statistical difference in overall mortality was found in studies defining ET as that done within 7 (837 patients; 22.9% vs. 35.0%; OR 0.64, 95% CI 0.35–1.15, p = 0.14; I2 = 19%) or 14 (2188 patients; 33.3% vs. 29.6%; OR 1.16, 95% CI 0.84–1.60, p = 0.37; I2 = 39%) days (Additional file 5: Figures S12–S13).

Discussion

By incorporating data from 14 studies involving 2371 tracheostomized COVID-19 patients, our systematic review and meta-analysis showed that ET was associated with improvement in 3 major clinical outcomes: duration of IMV, duration of ICU stay, and VAP. No differences were noted in overall mortality and time from tracheostomy to ventilator weaning between ET versus LT. Duration of sedation was reported by only one study, and hence, remains undetermined.

The timing of tracheostomy in ventilated COVID-19 patients has been the subject of debate [14, 42]. Our study demonstrated that, compared with LT, ET was associated with shorter durations of IMV and ICU stay. However, overall mortality rate was similar between patients who had ET and those who had LT. These findings align with the meta-analysis recently conducted by Chorath et al. in non-COVID-19 patients [43]. This has important implications for resource planning in a global pandemic, where the ventilator capacity is inadequate to meet heightened ventilator needs.

Evidence showed that VAP is a frequent complication among ventilated COVID-19 patients, which has a negative effect on outcomes [44,45,46]. Our findings indicate that ET may reduce the incidence of VAP. Given that VAP was a secondary outcome and that only two studies reported this outcome, we are fully cognizant that this outcome is speculative. Regardless of this shortcoming, several previous meta-analyses have reported that ET was associated with lower VAP rate in non-COVID-19 patients [43, 47, 48].

Although a previous monocentric study found that ET reduced duration of IMV, the reduction was specifically as a result of shortening the period from intubation to tracheostomy [49]. By contrast, a recent multicentric study included patients from the previous study showed that ET also reduced weaning time [37]. Our meta-analysis included this multicentric study and found a trend that patients undergoing ET had shorter duration of post-tracheostomy mechanical ventilation, although this was not statistically significant. The lack of statistical significance highlights the indication of the tracheotomy was a key factor for reducing the overall length of time required on IMV.

During the pandemic, the challenges of the logistics of patient selection, tracheostomy insertion and subsequent management, and health care worker safety may make LT seem more feasible in COVID-19 patients. One critique of ET is that ET will only free up ICU capacity in patients requiring prolonged ventilation. That is to say, it is possible that LT might lead to a reduced tracheostomy exposure, either because death occurs before tracheostomy is performed or because pulmonary recovery obviates the need for tracheostomy. However, our findings for the beneficial effect of ET on several clinical outcomes, such as duration of IMV, duration of ICU stay, and incidence of VAP, might question the current strategy of delaying tracheostomy in COVID-19 patients.

This is the largest and most comprehensive meta-analysis to date examining tracheostomy timing in patients with COVID-19. Unlike a meta-analysis included studies published before March 4, 2021 [50], we observed a decrease in time to ventilation weaning when patients underwent ET. This is most likely attributable to the addition of several studies after March 4, 2021. Another meta-analysis has also failed to show that ET improves the rate of overall mortality, but this meta-analysis was limited due to the small number of patients investigated [51].

Our meta-analysis also has limitations. First, our work is based on data from observational studies, which may suffer from residual confounding. Ideally, the outcomes of ET versus LT in ventilated COVID-19 patients should be evaluated in prospective, randomized trials; however, such studies are difficult to perform under pandemic conditions [52]. Second, as concerns the outcomes of the duration of IMV and ICU stay, we noted substantial statistical heterogeneity. Nevertheless, our sensitivity analyses have also found that ET reduced duration of IMV and ICU stay. Third, there is difference in definitions of early and late tracheostomy. This may introduce heterogeneity and could affect the results. We have tried to overcome this heterogeneity through doing a subgroup analysis according to the methodology of defining ET. Fourth, there has important progress in the management of patients with COVID-19 since the first wave of pandemic, which may attenuate the benefits of ET. We have not undertaken subgroup analyses between waves due to the majority of studies being performed during the first wave. Fifth, the heterogeneity in the treatment strategies employed by various authors (e.g., the introduction of steroids, etc.) could not be controlled for. Finally, only one study used ventilator-free and ICU-free days as composite measures of the effectiveness of ET in freeing up ICU resources [21]. Therefore, we did not choose ventilator-free and ICU-free days as the primary outcomes.

Conclusions

In summary, the findings from this meta-analysis suggest that ET in COVID-19 patients may reduce duration of IMV and ICU stay without modifying the mortality rate. This has implications for alleviating critical care capacity strain during the COVID-19 pandemic. Considering that tracheostomy is an aerosol-generating procedure, future studies are required to establish the role of timing in optimizing outcomes from tracheostomy and minimizing the risk of infection among health care workers.