British Journal of Anaesthesia 114 (3): 396–405 (2015)

Advance Access publication 22 December 2014 . doi:10.1093/bja/aeu440

Effect of early tracheostomy on resource utilization and

clinical outcomes in critically ill patients: meta-analysis
of randomized controlled trials
EDITOR’S CHOICE

T. Szakmany 1,2*, P. Russell 1, A. R. Wilkes 1 and J. E. Hall1

1
Cardiff Institute of Infection and Immunity, Cardiff University, UHW Main Building, Heath Park, Cardiff CF14 4XN, UK
2
Cwm Taf UHB, Royal Glamorgan Hospital, Llantrisant, UK
* Corresponding author. E-mail: szakmanyt1@cardiff.ac.uk

Background. Early tracheostomy may decrease the duration of mechanical ventilation,

Editor’s key points
† It is unclear whether early
tracheostomy in acutely
ill, ventilator-dependent
patients reduces costs
and complications.
† Small trials are unlikely to
reliably estimate relative
benefits and risks of
tracheostomy.
† This study identifies a
clear benefit of reduced
duration of sedation.
† A policy of early
tracheostomy will
however increase the
number of procedures
being undertaken.
sedation exposure, and intensive care stay, possibly resulting in improved clinical outcomes,
but the evidence is conflicting.
Methods. Systematic review and meta-analysis of randomized trials in patients allocated to
tracheostomy within 10 days of start of mechanical ventilation was compared with placement
of tracheostomy after 10 days if still required. Medline, EMBASE, the Cochrane Controlled
Clinical Trials Register, and Google Scholar were searched for eligible trials. The co-primary
outcomes were mortality within 60 days, and duration of mechanical ventilation, sedation,
and intensive care unit stay. Secondary outcomes were the number of tracheostomy
procedures performed, and incidence of ventilator-associated pneumonia (VAP). Outcomes are
described as relative risk or weighted mean difference with 95% confidence intervals.
Results. Of note, 4482 publications were identified and 14 trials enrolling 2406 patients were
included. Tracheostomy within 10 days was not associated with any difference in mortality
[risk ratio (RR): 0.93 (0.83–1.05)]. There were no differences in duration of mechanical
ventilation [20.19 days (21.13–0.75)], intensive care stay [20.83 days (22.05–0.40)], or
incidence of VAP. However, duration of sedation was reduced in the early tracheostomy groups
[22.78 days (23.68 to 21.88)]. More tracheostomies were performed in patients randomly
assigned to receive early tracheostomy [RR: 2.53 (1.18–5.40)].
Conclusion. We found no evidence that early (within 10 days) tracheostomy reduced mortality,
duration of mechanical ventilation, intensive care stay, or VAP. Early tracheostomy leads to
more procedures and a shorter duration of sedation.
Keywords: complications; early medical intervention; survival; tracheostomy; ventilator-
associated pneumonia
Accepted for publication: 29 September 2014

Tracheostomy is commonly performed in critically ill patients
with the objective of increasing comfort and shortening the dur-
ation of sedation, mechanical ventilation, and intensive care
stay.1 However, the evidence to confirm this benefit is unclear.2
The alternative is prolonged tracheal intubation which carries
the riskof respiratory tract injuryand othercomplications includ-
ing ventilator-associated pneumonia (VAP) and sinusitis.3 4
Tracheostomy is associated with procedure-related complica-
tions, including bleeding, hypoxia, oesophageal rupture, tra-
cheal stenosis, tracheal granulomas, and death.2 5 – 8 There
has been a significant increase in the utilization of tracheosto-
mies especially since the introduction of bedside percutaneous
tracheostomy in the mid-1980s.9 – 11 It has been estimated that
up to one-third of patients who undergo mechanical ventilation
in the intensive care unit (ICU) will undergo tracheostomy.10 11
As the percutaneous technique has become widely available,
the earlier use of tracheostomy has become commonplace.10 11
Consequently, there is ongoing debate about the benefits of
early tracheostomy. The objective of this study was to summar-
ize the available evidence through a systematic review and
meta-analysis. Specifically, we wished to confirm the effects
of tracheostomy within 10 days on critical care resource utiliza-
tion and short-term mortality compared with late tracheos-
tomy or prolonged intubation.
Methods
Search strategy
We followed the Preferred Reporting Items for Systematic Reviews
and Meta-Analyses recommendations for this meta-analysis.

& The Author 2014. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved.
For Permissions, please email: journals.permissions@oup.com
Timing of tracheostomy in critical illness
Two authors (P.R. and T.S.) independently performed the
electronic searches.
We searched the following databases: Cochrane Central
Register of Clinical Trials (CENTRAL) (The Cochrane Library
2013, Issue 13); MEDLINE (January 1950 to February 2014);
EMBASE (January 1980 to February 2014); CINAHL (1982
to February 2014); the NHS Trusts Clinical Trials Register and
Current Controlled Trials (www.controlled-trials.com); LILACS;
KoreaMED; MEDCARIB; INDMED; PANTELEIMON; Ingenta;
ISI Web of Knowledge and the National Trials Register to
identify all relevant randomized controlled trials available for
review using the strategy detailed in Supplementary material,
Appendix S1. We searched the bibliographies of reports of ran-
domized trials and any identified reviews. Ongoing clinical
trials were identified from the clinicaltrials.gov website, and
additional studies of interest were found through Internet
searches on Google Scholar and hand searches of bibliograph-
ies. We identified relevant studies initially by title then by ab-
stract and finally by full text. All studies in human beings that
were published in full text, abstract, or poster form were eligible
for inclusion, with no restrictions on publication date, lan-
guage, or status. The authors resolved any discrepancies by
discussion, if necessary.
Selection criteria
We included randomized clinical trials conducted in adult crit-
ically ill patients expected to require prolonged mechanical
ventilation of between 24 h and 21 consecutive days, for
more than 6 h per day. We included trials where one of the
groups received early tracheostomy, this must be carried out
within 10 days of mechanical ventilation; the alternative is pro-
longed tracheal intubation with the potential for a tracheos-
tomy to be placed after 10 days.
Unmasked quality assessment on the selected published
studies (not abstract reports) was carried out by two investiga-
tors, (T.S., P.R.) on composite aspects of study quality. To draw
conclusions about the overall risk of bias for an outcome it
was necessary to evaluate the trials for major sources of bias,
also defined as domains (random sequence generation, alloca-
tion concealment, blinding, incomplete outcome data, select-
ive outcome reporting, and other sources of bias).
Data extraction
Data extracted for each eligible study included: author; year of
publication; number of subjects; timing of tracheostomy;
number of procedures performed in each group; primary and
other study outcomes; commercial support; mortality within
60 days; mortality at the longest reported follow-up, incidence
of VAP; incidence of complications of procedure (where
reported).
If sufficient studies were identified we constructed funnel
plots (trial effect vs standard error) to assess for possible pub-
lication bias, expressed by asymmetry.12 In the case of asym-
metry we chose to apply the Arcsine–Thompson test, as
proposed by Rücker and colleagues.12 In case of publication
BJA
bias, we have repeated the analysis by removing the affected
trial from the analysis.
Data collection and evaluation
Two authors (P.R., T.S.) independently extracted data (as far as
possible) on the basis of an intention-to-treat analysis and
entered all data independently into Review Manager
(RevMan 5.3.1.) after checking for differences.
We used the Mantel–Haenszel model to calculate pooled
risk ratios (RRs) and 95% confidence intervals (CIs) with
random effects model or fixed-effect model depending on
the presence or absence of statistical heterogeneity, respect-
ively. Heterogeneity across studies was measured by I2 statis-
tics examining the percentage of heterogeneity because of
variation between studies (0% suggest no heterogeneity;
a value between 0 and 25% suggests very low heterogeneity;
a value between 25 and 50% suggests low heterogeneity; a
value between 50 and 75% suggests moderate heterogeneity;
a value of .75% suggests high heterogeneity). When I2 was
.50% we applied the random effects model as described
before. The mean difference for continuous data was analysed
using the inverse variance method.
Outcome measures
The co-primary outcomes were short-term mortality within 60
days, duration of mechanical ventilation, duration of sedation,
and duration of intensive care stay. Secondary outcomes were
the number of tracheostomy procedures performed, ventilator-
associated pneumonia and mortality at longest follow-up.
Results
We identified 4482 potential studies in the initial electronic
search. No additional studies were identified after screening
of reference lists of potentially eligible studies and previously
published systematic reviews (Fig. 1). We included 14 published
trials conducted between 1976 and 2011 and including 2406
patients.13 – 26 A detailed description of each trial included
can be found in Table 1. Combining the data from the studies
showed no significant difference in the relative risk of short-
term (up to 60-days) mortality between the groups: 356/
1180 (30.2%) deaths in the early tracheostomy vs 391/1226
(31.9%) deaths in the prolonged intubation group, RR: 0.93
(95% CI 0.83, 1.05; I2 ¼12%) (Fig. 2).
Early tracheostomy was not associated with any significant
difference in duration of intensive care stay, duration of mech-
anical ventilation or incidence of VAP (Figs 3–5). We found that
the duration of sedation was significantly shorter in the early
tracheostomy group (5 studies, 1425 patients, 22.78 days
95% CI: 23.68, 21.88) (Fig. 6).
There was no difference in the long-term outcome, which
was assessed at the longest reported time by the studies (14
studies, 2281 patients RR: 0.95 95% CI: 0.87, 1.03 I2 ¼0%)
(Fig. 7). Where data were available, we analysed the number
of those patients randomized to an early tracheostomy treat-
ment arm and those who received this allocated treatment.
The tracheostomy utilization was significantly higher in the
397
BJA Szakmany et al.

bias and when we removed it from the analysis, the statistical

heterogeneity decreased sufficiently to allow us to use the
4480 records 2 additional
identified records
fixed-effect model.14 We believe this trial is indeed affected
through identified by publication bias, as we have identified several registered
database through other studies from the same period, which have never been pub-
searching sources lished in full.

109 records after duplicates Discussion

removed and search limited
to randomized controlled
The principal finding of this analysis was that tracheostomy
trials within 10 days of onset of mechanical ventilation had no
effect on short-term mortality, duration of mechanical ventila-
tion, or duration of intensive care stay. Early tracheostomy was
109 records 89 records associated with a reduction in duration of sedation. Secondary
screened excluded outcomes were also similar in the two groups without signifi-
cant differences in long-term mortality and rate of VAP. Signifi-
cantly more tracheostomy procedures were carried out in the
6 full-text early tracheostomy cohort.
articles excluded: There were 14 studies with more than 2400 participants eli-
2 retrospective gible for inclusion in this meta-analysis. One study was identified
cohort analysis by our funnel plot assessment as a potential source of publica-
2 quasi tion bias, and this had been directly sponsored by industry.14
randomized trial
We identified several registered randomized clinical trials from
1 observational the same period, which had not been completed or published
study
in full, supporting this suggestion. Most of the included trials
1 not meeting were small, single centre and vulnerable to bias. Due to the
20 full-text inclusion criteria
articles assessed for early nature of the intervention appropriate blinding is impossible
for eligibility tracheostomy and this creates a risk of bias even in the large multicentre trials.
We found that short-term (up to 60 days) and long-term (up
to 2 yr) mortality was not significantly different between the two
14 studies groups. The funnel plot suggested publication bias caused by the
included in Bosel study, however even when this was included in the ana-
qualitative
synthesis lysis, we found the result was robust across the studies
without significant heterogeneity between the trials (Figs 1
and 2).25 All deaths in the Bosel study were attributed to the pro-
14 studies gression of the neurological sequelae in both groups and the
included in authors have concluded that the timing of tracheostomy is un-
quantitative likely to have influenced outcome.25
synthesis Although there was some heterogeneity among trial results
(meta-analysis)
with respect to mortality, all trials included critically ill patients
where various pathophysiological disturbances led to a com-
Fig 1 Preferred PRISMA flow diagram detailing search strategy and mon pathway of prolonged respiratory failure. Consequently,
identification of studies used in data synthesis. we believe that there is a good biological rationale to perform
a broad meta-analysis, which also considerably increases the
early tracheostomy group (RR: 2.53 95% CI: 1.18, 5.40) (Fig. 8). generalizability and usefulness of the review.28
This was because in the late group either the patient was Based on the results, it is difficult to explain how the timing
weaned from mechanical ventilation successfully or the of an intervention aimed at reduction of potential long-term
patient died before a tracheostomy was inserted. complications of intensive care would be able to directly influ-
Funnel plot analysis suggested significant publication bias ence mortality.14 19
in the primary outcome in one of the trials.25 However, upon It has been postulated that early tracheostomy may reduce
closer examination of the results as suggested by Sterne and ICU resource utilization, namely length of mechanical ventila-
colleagues27 we determined this was due to methodological tion and facilitate earlier discharge to the ward or in the USA, to
flaws in the Bosel trial, rather than publication bias. long-term care facility, hence reducing length of ICU and acute
When examining the secondary outcomes, funnel plot ana- hospital stay.14 29
lysis suggested reporting bias in duration of mechanical venti- Perhaps surprisingly, we could not find this effect in our
lation, duration of intensive care stay and incidence of VAP. In meta-analysis. The duration of mechanical ventilation was
all cases the Rumbak trial proved to be the source of publication identical in the two groups as was length of ICU stay.

398
Timing of tracheostomy in critical illness BJA

Table 1 Characteristics of included studies

Description
Barquist and colleagues Methods Prospective, randomized trial. Early tracheostomy group was allocated to receive a tracheostomy before
(2006)20 Day 8 of mechanical ventilation. The control group was allocated to receive a tracheostomy after Day 28.
Eligible patients were randomized before Day 8
Participants Adult ventilator dependent patients admitted to a trauma centre. They must have been intubated for at
least 3 days, 7 days after admission to the ICU, to be included
Interventions The insertion of an early tracheostomy (as opposed to delayed intubation and a late tracheostomy)
Outcomes Days on mechanical ventilation, days on mechanical ventilation after tracheostomy, length of ICU stay,
length of ICU stay after tracheostomy, incidence of VAP, tracheostomy complications, survival
Blot and colleagues Methods Prospective, randomized trial
(2008)16 Participants ICU patients ventilated for less than 4 days, but expected to be ventilated more than 7 days
Interventions The insertion of a tracheostomy before the fourth calendar day of mechanical ventilation, referred to as
‘early’ in this study. Late tracheostomy was one placed after 14 days of mechanical ventilation
Outcomes The primary outcome was 28-day mortality.
Secondary outcomes: duration of mechanical ventilation, the length of ICU stay and sedation
requirements, Day 60 and hospital mortality rates, laryngeal and tracheal complications, rate of VAP and
patient comfort
Bosel and colleagues Methods Prospective, randomized trial. Those allocated to early tracheostomy received the intervention within
(2012)25 3 days of intubation; those allocated to prolonged intubation had a tracheostomy between 7 and 14 days
if still clinically indicated
Participants Patients were 18 yr and over, had a diagnosis of non-traumatic intracerebral haemorrhage,
subarachnoid haemorrhage or acute ischaemic stroke and required mechanical ventilation for at least
14 days
Interventions The insertion of an early tracheostomy
Outcomes Mortality, length of ICU stay, duration of mechanical ventilation, type of mechanical ventilation needed,
weaning phases, level of sedation, complications, cost
Bouderka and colleagues Methods Randomized, controlled trial carried out in 1 Moroccan ICU
(2004)21 Participants A sample size of 62 head injury ICU patients with a GCS ,8 on Day 5 of mechanical ventilation was
available for analysis
Interventions Early tracheostomy vs prolonged intubation. Early tracheostomy was carried out on the fifth or sixth day.
No mention of the possibility of a tracheostomy being carried out on a patient in the prolonged intubation
group
Outcomes Length of mechanical ventilation, incidence of pneumonia, mortality
Bylappa and colleagues Methods Prospective, randomized trial. An early tracheostomy was defined as a tracheostomy inserted between
(2011)17 Days 5 and 7 of mechanical ventilation; prolonged intubation was defined as prolonged tracheal
intubation with a tracheostomy inserted between Days 8 and 15 of mechanical ventilation.
Randomization occurred at Day 4 if a patient was expected to require mechanical ventilation for another
6 days
Participants ICU patients requiring prolonged mechanical ventilation for 10 days
Interventions The insertion of an early tracheostomy
Outcomes Duration of mechanical ventilation; complications; duration of hospital stay
El-Naggar and colleagues Methods Prospective, randomized trial. Randomization occurred on Day 3. Early tracheostomy was defined as a
(1976)23 tracheostomy inserted on Day 3. Prolonged intubation was defined as ventilation via a tracheal tube for
10 –11 days with the insertion of tracheostomy if clinically relevant
Participants Adults requiring prolonged mechanical ventilation
Interventions The insertion of an early tracheostomy
Outcomes The patient’s epidemiological variables; daily pulmonary functions, severity of respiratory infections, and
scores of post-intubation airway lesions
Rumbak and colleagues Methods Randomized, controlled trial
(2004)14 Participants ICU patients projected to need ventilation for .14 days
Interventions Early vs late tracheostomy. Early tracheostomy was carried out within 48 h of mechanical ventilation.
Late tracheostomy was carried out within 14 –16 days of mechanical ventilation
Outcomes Mortality, rate of VAP, length of ICU stay, duration of mechanical ventilation, duration of sedation,
duration of inotropic support, organisms causing pneumonia
Saboori and colleagues Methods Randomized controlled single centre trial
(2009)18 Participants ICU patients admitted after head trauma
Interventions Early tracheostomy on the fourth day of ICU admission, late tracheostomy on Day 14
Outcomes Mortality, rate of VAP, duration of mechanical ventilation, length of ICU stay
Saffle and colleagues Methods Prospective, randomized trial. Early tracheostomy was defined as a tracheostomy inserted within one
(2002)22 working day of randomization; prolonged intubation was defined as continued tracheal intubation with a
tracheostomy inserted on Day 14 if clinically indicated

Continued

399
BJA Szakmany et al.

Table 1 Continued

Description
Participants Adult patients who were hospitalized within 24 h of acute burns injury, requiring on-going mechanical
ventilation
Interventions The insertion of an early tracheostomy
Outcomes Mortality, duration of mechanical ventilation, length of stay
Sugerman and colleagues Methods Multicentre, prospective, randomized trial. Patients were anticipated that they would require mechanical
(1997)24 ventilation for at least 7 days. Early tracheostomy was defined as a tracheostomy inserted between
Days 3 and 5; prolonged intubation was defined as a tracheostomy inserted, if clinically relevant, after
10 –14 days
Participants Head trauma, non-head trauma, and critically ill non-trauma patients requiring prolonged mechanical
ventilation
Interventions The insertion of an early tracheostomy
Outcomes Mortality, length of stay on the ICU, incidence of VAP, evidence of short- or long-term pharyngeal,
laryngeal, or tracheal injury.
Terragni and colleagues Methods Randomized controlled trial performed in 12 Italian ICUs
(2010)13 Participants Patients requiring prolonged mechanical ventilation
Interventions Early tracheostomy vs late tracheostomy. Early tracheostomy was defined as one carried out within
6 – 8 days of mechanical ventilation. A late tracheostomy occurred after 15 days of mechanical
ventilation
Outcomes The primary outcome was the incidence of VAP.
The secondary outcomes were at Day 28: number of ventilator-free days, number of ICU-free days,
mortality
Trouillet and colleagues Methods Prospective randomized controlled single center trial
(2011)15 Participants Patients who still required mechanical ventilation 4 days after operation after cardiac surgery
Interventions Immediate tracheostomy (early tracheostomy group) vs prolonged mechanical ventilation with a
possibility of a tracheostomy after Day 15 of randomization
Outcomes Number of days alive and breathing without ventilatory support.
Secondary end points: number of ventilator-free Days at 28 and 90 days (based on data through 28 and
90 days); 28-, 60-, and 90-day mortality rates; duration of mechanical ventilation and length of ICU and
hospital stays, sedation-free days at Day 28, the number of tracheal prosthesis-free days at Day 60;
frequencies of unscheduled extubations, decannulations and reintubations or recannulations; 7-, 14-,
21-, and 28-day sequential organ function assessment scores in the ICU; duration of vasopressor and
renal replacement therapy, rate of VAP
Young and colleagues Methods Randomized clinical trial in 72 centres within the UK
(2013)19 Participants ICU patients expected to require seven more days of mechanical ventilation
Interventions Early vs late tracheostomy. An early tracheostomy was one placed within 4 days of mechanical
ventilation. A late tracheostomy was one placed on Day 10 or later if still clinically relevant
Outcomes The primary outcome was the incidence of mortality within the first 30 days of mechanical ventilation.
The secondary outcomes are: mortality at ICU and hospital discharge and at 1 and 2 yr, length of ICU stay,
antibiotic-free days
Zheng and colleagues Methods A prospective randomized controlled trial in a single Chinese ICU
(2012)26 Participants ICU patients estimated to require mechanical ventilation for more than 14 days
Interventions Early tracheostomy vs late tracheostomy. Early tracheostomy was performed on Day 3 of mechanical
ventilation. Late tracheostomy was carried out on Day 15 of mechanical ventilation if still required
Outcomes The primary outcome was the number of ventilator-free days at Day 28 after randomization.
The secondary outcomes were: sedation-free and ICU-free days, successful weaning and ICU discharge
rate, incidence of VAP at Day 28, mortality at 60 days

We could not find any effect of a protocolized weaning
process in our sensitivity analysis between the early tracheos-
tomy and prolonged intubation groups when we grouped
together only those studies where a weaning protocol was
used (data not shown). Our results suggest, that in the pres-
ence of a structured approach to weaning from mechanical
ventilation, the type of breathing tube is not a determinant of
outcome.
The duration of sedation was significantly shorter in the
early tracheostomy.
Less sedative use was a uniform finding among all studies,
which reported this outcome, however, the heterogeneity
between the trials was again statistically significant. This is a
concern and our data should be interpreted with caution. It
could partly be explained by the different sedation protocols
and pharmacological agents used in the various trials. Only
two of the trials described the use of a protocol driven sedation
and analgesia regime and the others only refer to guidelines for
sedative use. It is important to emphasize that there is a gulf
between perceived and actual sedation practices in ventilated
patients, with only a fraction of evidence-based recommenda-
tions currently used in the majority of ICUs.30 It has been
shown that nurse-led protocol driven sedation, analgesia and
delirium management can significantly reduce the

400
Timing of tracheostomy in critical illness BJA

Early tracheostomy Prolonged intubation Risk ratio Risk ratio

Study or Subgroup Events Total Events Total Weight M-H, Fixed, 95% CI Year M-H, Fixed, 95% CI
El-Naggar 1976 21 26 21 26 5.4% 1.00 (0.77, 1.30) 1976
Sugerman 1997 13 53 19 95 3.5% 1.23 (0.66, 2.28) 1997
Rumbak et al 2004 11 60 18 60 4.7% 0.61 (0.32, 1.18) 2000
Saffle et al 2002 4 21 6 23 1.5% 0.73 (0.24, 2.23) 2001
Bouderka et al 2004 12 31 7 31 1.8% 1.71 (0.78, 3.77) 2002
Blot et al 2008 12 61 15 62 3.9% 0.81 (0.42, 1.59) 2004
Barquist 2006 2 29 5 31 1.3% 0.43 (0.09, 2.03) 2004
Terragni et al 2010 65 209 62 210 16.0% 1.05 (0.79, 1.41) 2008
Saboori, 2009 2 20 2 20 0.5% 1.00 (0.16, 6.42) 2008
Young et al 2013 168 451 180 448 46.9% 0.93 (0.79, 1.09) 2008
Trouillet 2011 30 109 28 107 7.3% 1.05 (0.68, 1.63) 2009
Bylappa 2011 0 22 0 22 Not estimable 2011
Zheng et al 2012 13 58 14 61 3.5% 0.98 (0.50, 1.90) 2011
Bosel 2012 3 30 14 30 3.6% 0.21 (0.07, 0.67) 2011

Total (95% CI) 1180 1226 100.0% 0.93 (0.83, 1.05)

Total events 356 391
Heterogeneity. Chi = 13.57, df = 12 (P = 0.33); I 2 = 12%
2

Test for overall effect: Z = 1.17 (P = 0.24) 0.1 0.2 0.5 1 2 5 10

Early tracheostomy Prolonged intubation

Fig 2 Forest plot of short-term mortality. Year: date of trial conducted.

Early tracheostomy Prolonged intubation Mean difference Mean difference

Study or subgroup Mean SD Total Mean SD Total Weight IV, Fixed, 95% CI Year IV, Fixed, 95% CI
Bouderka et al 2004 14.5 7.3 31 17.5 10.2 31 4.5% –3.00 (–7.42, 1.42) 2002
Young et al 2013 13 8.1 451 13.1 12 448 49.2% –0.10 (–1.44, 1.24) 2008
Terragni et al 2010 28 9.62 209 28 5.93 210 37.7% 0.00 (–1.53, 1.53) 2008
Saboori, 2009 20 58 20 28 15 20 0.1% –8.00 (–34.26, 18.26) 2008
Trouillet 2011 23.9 21.3 109 25.2 22.2 107 2.6% –1.60 (–7.40, 4.20) 2009
Bosel 2012 18 8.88 30 17 6.66 30 5.6% 1.00 (–2.97, 4.97) 2011
Zheng et al 2012 20 13 58 25 75 61 0.2% –5.00 (–24.12, 14.12) 2011

Total (95% CI) 908 907 100.0% –0.19 (–1.13, 0.75)

Heterogeneity. Chi2 = 2.79, df = 6 (P = 0.84); I 2 = 0%
Test for overall effect: Z = 0.40 (P = 0.69) –20 –10 0 10 20
Early tracheostomy Prolonged intubation

Fig 3 Forest plot of length of intensive care stay (days). Year: date of trial conducted.

Early tracheostomy Prolonged intubation Mean difference Mean difference

Study or subgroup Mean SD Total Mean SD Total Weight IV, Fixed, 95% CI Year IV, Fixed, 95% CI
Saffle et al 2002 35.5 20.62 21 31.4 24.94 23 0.8% 4.10 (–9.38, 17.58) 2001
Bouderka et al 2004 14.5 7.3 31 17.5 10.6 31 7.3% –3.00 (–7.53, 1.53) 2002
Barquist 2006 21.43 22.1 29 21.17 21 31 1.3% 0.26 (–10.66, 11.18) 2004
Blot et al 2008 14 13.78 61 16 13.53 62 6.5% –2.00 (–6.83, 2.83) 2004
Saboori, 2006 8 9 20 12 25 20 1.1% –4.00 (–15.64, 7.64) 2008
Young et al 2013 13.6 12 451 15.2 14.4 448 50.1% –1.60 (–3.33, 0.13) 2008
Trouillet 2011 17.9 14.9 109 19.3 16.9 107 8.3% –1.40 (–5.65, 2.85) 2009
Bosel 2012 15 5.18 30 12 5.92 30 19.0% 3.00 (0.19, 5.81) 2011
Zheng et al 2012 18.43 10.9 58 20.62 17.2 61 5.7% –2.19 (–7.34, 2.96) 2011

Total (95%CI) 810 813 100.0% –0.83 (–2.05, 0.40)

Heterogeneity. Chi2 = 10.15, df = 8 (P = 0.25); I 2 = 21%
Test for overall effect: Z = 1.32 (P = 0.19) –20 –10 0 10 20
Early tracheostomy Prolonged intubation

Fig 4 Forest plot of duration of mechanical ventilation (days). Year: date of trial conducted.

unnecessary over sedation and improve patient comfort
regardless of the type of breathing tube used for mechanical
ventilation.31 Importantly, long-term cognitive impairment
does not seem to be associated with reduction of sedative use
and it is uncertain if reduced sedation is beneficial in reducing
the incidence and severity of ICU delirium.31 32 Based on our
results it is impossible to ascertain if early tracheostomy is
indeed the main reason for the reduction of sedative exposure,
or whether other strategies could achieve the same results.
We have investigated whether important complications
usually associated with a worse outcome, for example VAP
rates, are different between the two groups. Randomized con-
trolled trials conducted at the beginning of the millennium and
retrospective case–control studies suggested that early
tracheostomy could reduce the rate of VAP.14 33 34
According to our analysis, the incidence of VAP was similar in
the two groups, but the heterogeneity between the trials was
statistically significant (Fig. 5). This heterogeneity could be

401
BJA Szakmany et al.

Early tracheostomy Prolonged intubation Risk ratio Risk ratio

Study or Subgroup Events Total Events Total Weight M-H, Random, 95% CI Year M-H, Random, 95% CI
Sugerman 1997 25 53 34 92 11.2% 1.28 (0.86, 1.89) 1997
Saffle et al 2002 21 21 22 23 20.5% 1.04 (0.92, 1.18) 2001
Bouderka et al 2004 18 31 19 31 10.7% 0.95 (0.63, 1.43) 2002
Blot et al 2008 30 61 31 62 12.2% 0.98 (0.69, 1.40) 2004
Terragni et al 2010 30 209 44 210 10.3% 0.69 (0.45, 1.05) 2008
Saboori, 2009 12 20 19 20 11.8% 0.63 (0.44, 0.92) 2008
Trouillet 2011 50 109 47 107 14.3% 1.04 (0.78, 1.40) 2009
Zheng et al 2012 17 58 30 61 9.0% 0.60 (0.37, 0.96) 2011

Total (95% CI) 562 606 100.0% 0.90 (0.75, 1.08)

Total events 203 246
Heterogeneity. Tau2 = 0.04; Chi2 = 18.20, df = 7 (P = 0.01); I 2 = 62%
Test for overall effect: Z = 1.12 (P = 0.26) 0.2 0.5 1 2 5
Early tracheostomy Prolonged intubation

Fig 5 Forest plot of incidence of VAP. Year: date of trial conducted.

Early tracheostomy Prolonged intubation Mean difference Mean difference

Study or Subgroup Mean SD Total Mean SD Total Weight IV, Random, 95% CI Year IV, Random, 95% CI
Blot et al 2008 12.5 6.75 61 13.75 6.75 62 10.1% –1.25 (–3.64, 1.14) 2004
Young et al 2013 5 4.44 455 7 5.18 454 29.5% –2.00 (–2.63, –1.37) 2008
Trouillet 2011 6.4 5.9 109 9.6 7.3 107 14.8% –3.20 (–4.97, –1.43) 2009
Bosel 2012 7.19 3.4 29 11 4 29 13.5% –3.81 (–5.72, –1.90) 2011
Zheng et al 2012 7.16 0.8 58 10.5 1.4 61 32.2% –3.34 (–3.75, –2.93) 2011

Total (95% CI) 712 713 100.0% –2.78 (–3.68, –1.88)

Heterogeneity. Tau 2 = 0.61; Chi 2 = 15.14, df = 4 (P = 0.004); I 2 = 74%
Test for overall effect: Z = 6.05 (P < 0.00001) –10 –5 0 5 10
Early tracheostomy Prolonged intubation

Fig 6 Forest plot of duration of sedation (days). Year: date of trial conducted.

Early tracheostomy Prolonged intubation Risk ratio Risk ratio

Study or Subgroup Events Total Events Total Weight M-H, Fixed, 95% CI Year M-H, Fixed, 95% CI
El-Naggar 1976 21 26 21 26 4.0% 1.00 (0.77,1.30) 1976
Sugerman 1997 13 53 19 95 2.6% 1.23 (0.66, 2.28) 1997
Saffle et al 2002 4 21 6 23 1.1% 0.73 (0.24, 2.23) 2001
Bouderka et al 2004 12 31 7 31 1.3% 1.71 (0.78, 3.77) 2002
Blot et al 2008 16 61 15 62 2.9% 1.08 (0.59, 1.99) 2004
Barquist 2006 2 29 5 31 0.9% 0.43 (0.09, 2.03) 2004
Saboori, 2009 2 20 2 20 0.4% 1.00 (0.16, 6.42) 2008
Young et al 2013 230 451 238 443 46.2% 0.95 (0.84, 1.08) 2008
Terragni et al 2010 137 209 147 210 28.2% 0.94 (0.82, 1.07) 2008
Trouillet 2011 33 109 32 107 6.2% 1.01 (0.67, 1.52) 2009
Bylappa 2011 0 22 0 22 Not estimable 2011
Bosel 2012 8 30 18 30 3.5% 0.44 (0.23, 0.86) 2011
Zheng et al 2012 13 58 14 61 2.6% 0.98 (0.50, 1.90) 2011
Total (95% CI) 1120 1161 100.0% 0.95 (0.87, 1.03)
Total events 491 524
Heterogeneity. Chi2 = 9.57, df = 11 (P = 0.57); I 2 = 0%
Test for overall effect: Z = 1.20 (P = 0.23) 0.1 0.2 0.5 1 2 5 10
Early tracheostomy Prolonged intubation

Fig 7 Forest plot of long-term mortality. Year: date of trial conducted.

explained by the very different tools and definitions used to de-
scribe VAP across the studies. There was insufficient patient
level data available to use a standardized VAP definition
across the studies. It is also unclear from the published trials
what, if any, evidence-based procedures were used to try to
prevent VAP. It has been shown that the systematic adoption
of these methods could significantly reduce the rate of VAP, re-
gardless of the type of tube used for mechanical ventilation.35
The results were too diverse for us to be able to group the
studies according to their chosen VAP definition, and conse-
quently this result should be interpreted with caution.
Tracheostomy utilization was significantly higher in the
early tracheostomy compared with the prolonged intubation
group. Only 417 of the 1214 patients randomized into the pro-
longed intubation group eventually received tracheostomy
(Fig. 8), as opposed to 1027 of the 1166 patients in the early
tracheostomy group. While in the early tracheostomy group
the main reason for not performing the procedure was cardio-
respiratory instability, in the prolonged intubation group in
most cases this was because either the patient was weaned
from mechanical ventilation successfully or the patient died
before a tracheostomy was inserted and only in a few cases

402
Timing of tracheostomy in critical illness BJA

Early tracheostomy Prolonged intubation Risk ratio Risk ratio

Study or Subgroup Events Total Events Total Weight M-H, Random, 95% CI Year M-H, Random, 95% CI
Sugerman 1997 53 53 23 95 8.7% 4.05 (2.85, 5.76) 1997
Rumbak et al 2004 60 60 50 60 8.8% 1.20 (1.07, 1.35) 2000
Saffle et al 2002 20 21 17 23 8.7% 1.29 (0.99, 1.67) 2001
Bouderka et al 2004 31 31 0 31 4.1% 63.00 (4.03, 985.89) 2002
Blot et al 2008 60 61 16 62 8.6% 3.81 (2.50, 5.82) 2004
Barquist 2006 27 29 11 31 8.5% 2.62 (1.62, 4.26) 2004
Terragni et al 2010 145 209 119 210 8.8% 1.22 (1.06, 1.42) 2008
Young et al 2013 385 455 33 454 8.7% 11.64 (8.36, 16.21) 2008
Trouillet 2011 109 109 29 107 8.7% 3.64 (2.68, 4.96) 2009
Koch 2012 50 50 50 50 8.8% 1.00 (0.96, 1.04) 2010
Zheng et al 2012 58 58 51 61 8.8% 1.19 (1.06, 1.34) 2011
Bosel 2012 29 30 18 30 8.7% 1.61 (1.19, 2.17) 2011

Total (95% CI) 1166 1214 100.0% 2.53 (1.18, 5.40)

Total events 1027 417
2 2 2
Heterogeneity. Tau = 1.70; Chi = 2815.13, df = 11 (P < 0.00001); I = 100%
Test for overall effect: Z = 2.39 (P = 0.02) 0.001 0.1 1 10 1000
Prolonged intubation Early tracheostomy

Fig 8 Forest plot of tracheostomy procedures performed. Year: date of trial conducted.

because there was an unintended violation of the study proto-
col. Despite this significant difference in the rate of procedure,
the overall outcome and resource utilization was similar in both
groups (Fig. 2). This suggests that patient outcomes are the
same in the prolonged intubation group without the added
risk of performing a surgical procedure early in their clinical
course. The implications of our findings are significant: out of
100 mechanically ventilated patients on an ICU which uses
an early tracheostomy policy, routinely inserting tracheosto-
mies before Day 10 of mechanical ventilation means that 54
more patients could receive a tracheostomy without any real
benefit in outcome. This excess activity could result in immedi-
ate, short- and long-term complications. Recent studies found
that percutaneous dilatational tracheostomy procedures carry
a 0.17–0.6% chance of fatal outcome.5 7 As this is a common
procedure on ICUs on the western hemisphere, the potential
of reducing harm by adopting a more conservative approach
to management is considerable. While the incidence of imme-
diate complications such as bleeding, pneumothorax or death
attributed to the procedure was similar in both groups, there
were only occasional reports on long-term functional out-
comes in terms of incidence of tracheal stenosis, tracheomala-
cia, or difficulty in swallowing.19 36 Incidence rates for
tracheostomy related short-term complications such as desat-
uration, bleeding, hypotension, and need for increased ventila-
tor support were similar at around 17% in both groups (data
not shown), hence we postulate that the unwarranted proce-
dures could also lead to avoidable harm.
It remains possible, that early tracheostomy is associated
with undetected harm or benefit in the general ICU population.
We believe that our approach offers significant advantages
over previously published work investigating the effects of
the timing of the procedure in the critically ill.
Compared with the Cochrane meta-analysis of Silva and col-
leagues2 published in 2012, we have included 14 trials, 5 of
which were published since the last date of the literature
search of the Silva study, with the inclusion of an additional
1388 patients.15 17 19 25 26 37 Silva and colleagues2 only summar-
ized four studies, including one, which had a quasi-randomized
design.38 This meta-analysis only examined whether early vs
late tracheostomy is associated with better outcomes, hence
excluded a number of clinically important studies, which com-
pared early tracheostomy with prolonged intubation. Because
of the restrictive nature of their search strategy it is not surpris-
ing that Silva and colleagues2 found insufficient data to success-
fully evaluate the procedure.2
Similarly, the very recent meta-analysis by Huang and col-
leagues39 only included nine studies. Compared with their ana-
lysis we have included five more studies including the very
recently concluded Bosel trial.25 The main findings of Huang
and colleagues39 are very similar to ours in terms of short-term
mortality, length of mechanical ventilation and length of ICU
stay. Importantly, they have also noted that the Rumbak
study was a source of statistical heterogeneity, but contrary
to our assessment, they did not find it to be a source of publica-
tion bias.39
Our meta-analysis has wider inclusion criteria with studies
representing the whole critical care spectrum, does not have lan-
guage or geographic restrictions and investigates more diverse
clinical outcomes including length of sedation. Our principle
aim was to investigate if there is a difference between outcomes
if early tracheostomy is performed as opposed to prolonged in-
tubation with a possibility of late tracheostomy. This question
more closely describes the dilemma encountered in usual critical
care practice and we believe our results are generalizable over a
wide range of clinical scenarios. The larger sample size enables
us to present clinically important findings, which can direct clin-
ical practice and further research.
Our meta-analysis has several limitations. First, as a
meta-analysis our research is retrospective and subject to
the methodological soundness of the individual studies. We
have tried to keep the probability of bias to a minimum by
developing a detailed protocol a priori, carrying out a thorough
search for published and unpublished data, and using explicit
criteria for study selection, data collection, and data analysis.
As a result, we consider that our robust approach has resulted
in recommendations directly applicable to clinical practice.
Secondly, our review includes trials from 1976 to 2012. There
has been an enormous change in clinical practice during this
period, which could account for the negative findings. However

403
BJA
when we grouped the trials according to their recruitment
period, we did not find any significant differences in the results
between trials conducted before or after the millennium (data
not shown).
Thirdly, there is little guidance on the prediction of prolonged
mechanical ventilation and the timing of tracheostomy inser-
tion is based on this assessment. The clinical judgement of the
attending clinician necessary to provide an estimate for the
length of mechanical ventilation carries the risk of selection
bias. Overall, all of the included studies have different definitions
of early tracheostomy and prolonged mechanical ventilation.
Consequently, we can only provide data on the safetyand effect-
iveness of early tracheostomy on reduction of mortality com-
pared with standard treatment.
It is clear that continued research is needed to find appropri-
ate tools to predict the duration of mechanical ventilation on
the ICU.
Future research should be aimed at standardizing the defini-
tions of early tracheostomy and examining if it would be bene-
ficial in certain patient groups. The safety and late complication
rates of tracheostomy are poorly understood and further
efforts should be directed to examine the wider socio-economic
consequences of the procedure.
Conclusions
Despite hypothesized and plausible clinical benefits, our ana-
lysis suggests that early tracheostomy does not carry any mor-
tality advantage in the heterogeneous patient population
included in this work. According to our data, early tracheos-
tomy does not help to reduce length of ICU stay or incidence
of VAP. There appears to be a reduction in the duration of seda-
tive use when performing early tracheostomy, although this is
not accompanied by a reduction in duration of mechanical ven-
tilation. Our results suggest that the use of early tracheostomy
leads to unnecessarily high procedural rate with associated
increased morbidity and possibly financial cost. This leads us
to suggest that tracheostomy before Day 10 of mechanical
ventilation should be avoided. Further research with adequately
powered and methodologically sound clinical trials should
address the questions if any particular subgroups of critically
ill patients would benefit from the procedure and to under-
stand the longer term effects of the intervention.
Supplementary material
Supplementary material is available at British Journal of
Anaesthesia online.
Authors’ contributions
T.S.: study design, data collection, data analysis and interpret-
ation of the results, and writing of the paper; P.R.: study design,
data collection, data analysis, and writing of the first draft of
the paper. A.R.W.: data analysis, interpretation of the results,
and writing of the first draft of the paper. J.E.H: study design, in-
terpretation of the results, and writing of the paper.
404
Szakmany et al.
Declaration of interest
None declared.
Funding
There was no funding source for this study. T.S. is a recipient of a
Clinical Research Fellowship from the National Institute for
Social Care and Health Research, Academic Health Science Col-
laboration scheme (2011 –2014).
References
1 Freeman BD, Morris PE. Tracheostomy practice in adults with acute
respiratory failure. Crit Care Med 2012; 40: 2890–6
2 Gomes Silva BN, Andriolo RB, Saconato H, Atallah AN, Valente O.
Early versus late tracheostomy for critically ill patients. Cochrane
Database Syst Rev 2012; 3: CD007271
3 Fernandez JF, Levine SM, Restrepo MI. Technologic advances in
endotracheal tubes for prevention of ventilator-associated pneu-
monia. Chest 2012; 142: 231– 8
4 Bishop MJ, Weymuller EA, Fink BR. Laryngeal effects of prolonged in-
tubation. Anesth Analg 1984; 63: 335– 42
5 Simon M, Metschke M, Braune SA, Püschel K, Kluge S. Death after
percutaneous dilatational tracheostomy: a systematic review and
analysis of risk factors. Crit Care 2013; 17: R258
6 Carrer S, Basilico S, Rossi S, Bosu A, Bernorio S, Vaghi GM. Outcomes
of percutaneous tracheostomy. Minerva Anestesiol 2009; 75:
607– 15
7 Gilbey P. Fatal complications of percutaneous dilatational
tracheostomy. Am J Otolaryngol 2012; 33: 770–3
8 Susarla SM, Peacock ZS, Alam HB. Percutaneous dilatational
tracheostomy: review of technique and evidence for its use. J Oral
Maxillofac Surg 2012; 70: 74– 82
9 Ciaglia P, Firsching R, Syniec C. Elective percutaneous dilatational
tracheostomy. A new simple bedside procedure; preliminary
report. Chest 1985; 87: 715– 9
10 Veenith T, Ganeshamoorthy S, Standley T, Carter J, Young P. Inten-
sive care unit tracheostomy: a snapshot of UK practice. Int Arch Med
2008; 1: 21
11 Cox CE, Carson SS, Holmes GM, Howard A, Carey TS. Increase in
tracheostomy for prolonged mechanical ventilation in North Caro-
lina, 1993– 2002. Crit Care Med 2004; 32: 2219–26
12 Rücker G, Schwarzer G, Carpenter J. Arcsine test for publication bias
in meta-analyses with binary outcomes. Statist Med 2008; 27:
746– 63
13 Terragni PP, Antonelli M, Fumagalli R, et al. Early vs late tracheotomy
for prevention of pneumonia in mechanically ventilated adult ICU
patients. JAMA 2010; 303: 1483–9
14 Rumbak MJ, Newton M, Truncale T, Schwartz SW, Adams JW,
Hazard PB. A prospective, randomized, study comparing early per-
cutaneous dilational tracheotomy to prolonged translaryngeal in-
tubation (delayed tracheotomy) in critically ill medical patients.
Crit Care Med 2004; 32: 1689–94
15 Trouillet J-L, Luyt C-E, Guiguet M, et al. Early percutaneous tracheot-
omy versus prolonged intubation of mechanically ventilated
patients after cardiac surgery: a randomized trial. Ann Internal
Med 2011; 154: 373– 83
16 Blot F, Similowski T, Trouillet J-L, et al. Early tracheotomy versus pro-
longed endotracheal intubation in unselected severely ill ICU
patients. Intensive Care Med 2008; 34: 1779– 87
17 Bylappa K, Mohiyudin A, Wilma Delphine Silvia CR, Krishnamurthy D,
Pyarajan MS. A comparative study of early and late tracheostomy in
Timing of tracheostomy in critical illness
patients requiring prolonged tracheal intubation, 2011. Availa-
ble from: http://www.waent.org/archives/2011/Vol4-2/20111215-
Tracheostomy-Intubation/late-tracheoto my.htm (accessed 9
September 2013)
18 Saboori M, Esmaili Taramsari M, Hosseini B. The effects of early
tracheostomy on outcomes of patients with severe head injury.
J Isfahan Med Sch 2009; 27: 211 –6
19 Young D, Harrison DA, Cuthbertson BH, Rowan K. Effect of early vs
late tracheostomy placement on survival in patients receiving
mechanical ventilation. JAMA 2013; 309: 2121–9
20 Barquist ES, Amortegui J, Hallal A, et al. Tracheostomy in ventilator
dependent trauma patients: a prospective, randomized intention-
to-treat study. J Trauma 2006; 60: 91– 7
21 Bouderka M, Fakhir B, Bouaggad A, Hmamouchi B, Hamoudi D,
Harti A. Early tracheostomy versus prolonged endotracheal intub-
ation in severe head Injury. J Trauma 2004; 57: 251 –4
22 Saffle JR, Morris SE, Edelman L. Early tracheostomy does not
improve outcome in burn patients. J Burn Care Rehabil 2002; 23:
431– 8
23 El-Naggar M, Sadagopan S, Levine H, Kantor H, Collins VJ. Factors in-
fluencing choice between tracheostomy and prolonged translaryn-
geal intubation in acute respiratory failure: a prospective study.
Anesth Analg 1976; 55: 195 –201
24 Sugerman HJ, Wolfe L, Pasquale MD, et al. Multicenter, randomized,
prospective trial of early tracheostomy. J Trauma 1997; 43: 741–7
25 Bosel J, Schiller P, Hook Y, et al. Stroke-related early tracheostomy
versus prolonged orotracheal intubation in neurocritical care trial
(SETPOINT): a randomized pilot trial. Stroke 2012; 44: 21– 8
26 Zheng Y, Sui F, Chen XK, et al. Early versus late percutaneous dila-
tional tracheostomy in critically ill patients anticipated requiring
prolonged mechanical ventilation. Chin Med J (Engl) 2012; 125:
1925– 30
27 Sterne JAC, Sutton AJ, Ioannidis JPA, et al. Recommendations for
examining and interpreting funnel plot asymmetry in meta-analyses
of randomised controlled trials. Br Med J 2011; 343: d4002
BJA
28 Gotzsche PC. Why we need a broad perspective on meta-analysis. It
may be crucially important for patients. Br Med J 2000; 321: 585– 6
29 Griffiths J, Barber VS, Morgan L, Young JD. Systematic review and
meta-analysis of studies of the timing of tracheostomy in adult
patients undergoing artificial ventilation. Br Med J 2005; 330: 1243
30 Gill KV, Voils SA, Chenault GA, Brophy GM. Perceived versus actual
sedation practices in adult intensive care unit patients receiving
mechanical ventilation. Ann Pharmacother 2012; 46: 1331– 9
31 Mehta S, Burry L, Cook D, et al. Daily sedation interruption in mech-
anically ventilated critically ill patients cared for with a sedation
protocol: a randomized controlled trial. JAMA 2012; 308: 1985– 92
32 Pandharipande PP, Girard TD, Jackson JC, et al. Long-term cognitive
impairment after critical illness. N Engl J Med 2013; 369: 1306– 16
33 Bickenbach J, Fries M, Offermanns V, et al. Impact of early vs. late
tracheostomy on weaning: a retrospective analysis. Minerva
Anestesiol 2011; 77: 1176–83
34 Devarajan J, Vydyanathan A, Xu M, et al. Early tracheostomy is asso-
ciated with improved outcomes in patients who require prolonged
mechanical ventilation after cardiac surgery. J Am Coll Surg 2012;
214: 1008–16.e4
35 Caserta RA, Marra AR, Durão MS, et al. A program for sustained
improvement in preventing ventilator associated pneumonia in
an intensive care setting. BMC Infect Dis 2012; 12: 234
36 Norwood S, Vallina VL, Short K, Saigusa M, Fernandez LG,
McLarty JW. Incidence of tracheal stenosis and other late complica-
tions after percutaneous tracheostomy. Ann Surg 2000; 232:
233– 41
37 Koch T, Hecker B, Hecker A, et al. Early tracheostomy decreases ven-
tilation time but has no impact on mortality of intensive care
patients: a randomized study. Langenbecks Arch Surg 2012; 397:
1001– 8
38 Dunham CM, LaMonica C. Prolonged tracheal intubation in the
trauma patient. J Trauma 1984; 24: 120–4
39 Huang H, Li Y, Ariani F, Chen X, Lin J. Timing of tracheostomy in crit-
ically ill patients: a meta-analysis. PLoS One 2014; 9: e92981
Handling editor: P. S. Myles
405