liu et al 2015.pdf

8/16/2019 liu et al 2015.pdf

http://slidepdf.com/reader/full/liu-et-al-2015pdf 1/9

Systematic Review

Early versus Late Tracheostomy: ASystematic Review and Meta-Analysis

Otolaryngology–

Head and Neck Surgery

2015, Vol. 152(2) 219–227

American Academy of

Otolaryngology—Head and Neck

Surgery Foundation 2014

Reprints and permission:

sagepub.com/journalsPermissions.nav

DOI: 10.1177/0194599814561606

http://otojournal.org

C. Carrie Liu, MD1, Devon Livingstone, MD1,Elijah Dixon, MD, MSc2, and Joseph C. Dort, MD, MSc1,3

No sponsorships or competing interests have been disclosed for this article.

Abstract

Objective. To investigate whether early tracheostomy leadsto improved outcomes compared with late tracheostomy.

Data Sources. Ovid MEDLINE (including PubMed), Embase,

and the Cochrane Central Register of Controlled Trials.

Review Methods. A systematic search was performed of theabove-mentioned databases according to PRISMA guide-lines. Data were collected on the following outcomes of

interest: hospital mortality, intensive care unit length of stay, length of mechanical ventilation, incidence of pneu-monia, laryngotracheal injury, and sedation use. Analysiswas performed using the RevMan 5 software (CochraneCollaboration, Oxford, England).

Results. Eleven studies were included for analysis. There was asignificant decrease in the intensive care unit length of stay in

the early tracheostomy group (weighted mean difference,29.13 days; 95% confidence interval [CI], 217.55 to

20.70; P = .03). There was no significant difference in hos-pital mortality (relative risk, 0.84; 95% CI, 0.67 to 1.04; P =

.11). A pooled analysis was not performed for the incidenceof pneumonia or length of mechanical ventilation, secondaryto considerable heterogeneity among the studies. None of the studies reporting laryngotracheal outcomes found a signif-

icant difference between the early and late tracheostomygroups, whereas all 3 studies reporting sedation use found asignificant decrease in the early tracheostomy group.

Conclusion. Early tracheostomy performed within 7 days of intuba-tion was associated with a decrease in intensive care unit length of

stay. No difference was found in hospital mortality. Insufficient datacurrently exist to make conclusions about the effect of early tra-cheostomy on the incidence of pneumonia, length of mechanicalventilation, laryngotracheal injury, or sedation use.

Keywords

tracheostomy, tracheotomy, prolonged intubation, systema-tic review, meta-analysis

Received July 8, 2014; revised October 13, 2014; accepted November

6, 2014.

Tracheostomy is one of the most commonly performed

procedures in the critical care population.1 A common

indication for its performance is in the event of pro-

longed intubation and ventilation.2 Compared with endotracheal

intubation, the proposed advantages of tracheostomy are that it

facilitates weaning from mechanical ventilation, allows for

quicker resumption of speech and oral intake, and is more com-

fortable for the patient.1-4 Another proposed advantage of tra-cheostomy is that there is a lower risk of laryngeal injury. Rates

of early laryngeal injury following endotracheal intubation have

been reported to be as high as 94%.5 Long-term sequelae such

as granulomas and laryngeal stenosis range from 5% to 12%,5,6

with increased length of endotracheal intubation associated with

a higher incidence of stenosis.6 With the development of more

flexible tubes with low-pressure cuffs and improved tube man-

agement strategies, laryngeal injury and the related sequelae are

less common.4 Laryngeal injury, however, remains an important

risk as the course of intubation lengthens.7

The optimal timing of tracheostomy in critically ill patients

requiring prolonged mechanical ventilation is debated.Convention dictates that a tracheostomy may be performed if

the course of mechanical ventilation exceeds 14 days.7 A con-

sensus conference held in 1989 stated that patients who require

mechanical ventilation for up to 10 days may be endotrache-

ally intubated, whereas tracheostomy is favored if mechanical

ventilation is predicted to be greater than 21 days.2 The evi-

dence for this, however, is limited and a defined pathway for

A podcast for this article is available at

http://otojournal.org. Scan this code with

your phone to hear the podcast instantly.

1Division of Otolaryngology–Head and Neck Surgery, Department of

Surgery, University of Calgary, Alberta, Canada2Division of General Surgery, Department of Surgery, University of Calgary,

Alberta, Canada3Ohlson Research Initiative, University of Calgary, Alberta, Canada

This article was presented at the 2014 AAO-HNSF Annual Meeting & OTO

EXPO; September 21-24, 2014; Orlando, Florida.

Corresponding Author:

Joseph C. Dort, Division of Otolaryngology–Head and Neck Surgery,

Department of Surgery, Ohlson Research Initiative, University of Calgary,

HRIC 2A02, 3280 Hospital Dr. NW, Calgary, AB, T2N 4Z6, Canada.

Email: [email protected]

at GEORGIAN COURT UNIV on February 17, 2015oto.sagepub.comDownloaded from

http://oto.sagepub.com/




8/16/2019 liu et al 2015.pdf


the management of the airway in this population of patients

does not exist.7,8

Our goal in this study is to perform a systematic review

and meta-analysis of existing literature, investigating whether

early tracheostomy leads to improved outcomes compared

with late tracheostomy in critically ill patients.

Methods

Search Strategy and Study Selection

A systematic search was performed according to PRISMA

(Preferred Reporting Items for Systematic Reviews and

Meta-Analyses) guidelines by 2 reviewers (C.C.L. and D.L.).

The MEDLINE (including PubMed) and Embase databases

were searched using key terms tracheo$ and intubation,

whereby the symbol $ denotes the inclusion of all variations

of the term beginning with the stem tracheo (ie, to include

both tracheostomy and tracheotomy). The search terms tra-

cheo$ and intubation were each cross-searched with the

MeSH term time factors to identify studies that examine thetiming of tracheostomy and intubation. The Cochrane Central

Register of Controlled Trials (CENTRAL) database was also

searched using the key words tracheostomy and tracheotomy.

Studies were limited to those published over the past 30

years (January 1, 1983, through September 1, 2013), in the

English language, and involving human subjects. The biblio-

graphies of relevant studies were hand-searched to identify

any additional studies.

Title and abstract review was performed once the initial list

of studies was generated. The reviewers met prior to commen-

cing study selection to ensure consistency in the application of

the inclusion criteria. Studies were selected if they met the 4inclusion criteria: (1) the study was a randomized or quasi-

randomized trial in design, (2) participants were critically ill

patients requiring prolonged mechanical ventilation, (3) the

study examined the outcomes of early tracheostomy compared

with late tracheostomy, and (4) the outcomes included at least

one of the following: hospital mortality, length of intensive

care unit (ICU) stay, length of mechanical ventilation, inci-

dence of pneumonia, laryngotracheal injury, or sedation use.

A full-text review was performed of any remaining stud-

ies following the title and abstract review. Additional stud-

ies that did not meet inclusion criteria were excluded.

Data Collection

The outcomes of interest were hospital mortality, length of

mechanical ventilation, length of ICU stay, incidence of pneu-

monia, laryngotracheal injury, and sedation use. A detailed

data collection sheet was created based on the outcomes of

interest. This underwent multiple revisions by the reviewers

before data collection to ensure usability and consistency. Data

were collected independently by 3 reviewers (C.C.L., E.D.,

and J.D.). During the data collection process, the risk of bias

for each study was assessed. Various risk of bias scales exist,

and there is no one scale that is clearly validated.9-11 We

chose to use the criteria from the Cochrane Handbook for

Systematic Reviews of Interventions.12

Statistical Analysis

Relative risk (RR) with 95% confidence intervals (CIs) was

used to express differences for dichotomous variables.

Weighted mean differences (WMDs) with 95% CIs were

used for continuous variables. The I 2 statistic was used to

quantify heterogeneity, whereby I 2 of 0% to 40% might not

be important, 30% to 60% may indicate moderate heteroge-neity, 50% to 90% may indicate substantial heterogeneity,

and 75% to 100% indicates considerable heterogeneity.12

We only performed pooled analysis when the heterogeneity

was less than 60%. Data analysis was performed using a

random-effects model, given the varied patient populations

and study methodologies. A P value of \.05 was considered

significant. Where meta-analysis was performed, sensitivity

analysis was undertaken by removing individual studies

from the pooled results to assess the extent to which each

study contributed to the effect size and heterogeneity.

RevMan 5 software (Cochrane Collaboration, Oxford,

England) was used for all statistical analyses.

Subgroup analysis was performed to investigate whether

certain clinical or methodological characteristics of the

studies could account for the statistical heterogeneity seen

and whether these characteristics influenced the results of

the meta-analysis. Specifically, we performed subgroup

analysis based on the timing of the early tracheostomy (ie,

3 days, 4-5 days, 7-8 days), as studies differed in their

definitions of early and late tracheostomy. We also exam-

ined outcomes based on the etiology of acute illness of the

study population.

This study contains only data from the published literature,

and no patient data were used; therefore, institutional review

board and ethics committee approval was not required.

Results

A total of 3893 studies were identified following the initial

search. A title and abstract review excluded 3873 studies,

leaving 20 studies for full-text review. One additional study

was identified during the bibliography search of relevant

studies for a total of 21 studies. Following full-text review,

11 studies remained (Figure 1). Table 1 summarizes the

key features of the included studies, and Table 2 sum-

marizes the risk of bias assessment. The etiology of the crit-

ical illness of the study populations varied from medical,neurological, traumatic, to mixed. The timing of the early

tracheostomy differed significantly across studies, ranging

from 2 to 8 days. The definition of late tracheostomy also

varied, ranging from 6 to 28 days. In 2 studies, late tra-

cheostomies were performed during a time frame that fell

into most other studies’ definitions of early tracheost-

omy.13,14 This overlapping of groups may cause inaccurate

results in a meta-analysis, as the effects attributed to

patients in the late groups in these studies may negate simi-

lar effects attributed to patients in the early groups in other

studies. As such, these studies were not included in the

pooled analyses. No life-threatening complications were

reported in any of the studies.

220 Otolaryngology–Head and Neck Surgery 152(2)






8/16/2019 liu et al 2015.pdf


Incidence of Pneumonia

Eight studies reported the incidence of pneumonia as an out-

come.15-22 Two studies used the clinical pulmonary infec-

tion score (CPIS) to diagnose pneumonia.21,22 Four studies

used varying combinations of leukocyte count, the presence

of fevers, chest x-ray findings, and bronchial cultures to

make the diagnosis of pneumonia,

15,18-20

while 2 studies did not specify their method of diagnosis.16,17 The crude cumula-

tive incidence of pneumonia was 35.9% in the early tra-

cheostomy groups (185 of 515 patients) and 45.5% of

patients in the late tracheostomy groups (244 of 536 patients).

A pooled analysis was not performed secondary to significant

heterogeneity of the measured effects from the individual

studies. Figure 2 summarizes the results from each study

without an overall pooled analysis. Three studies reported a

significant decrease in the incidence of pneumonia in the

early tracheostomy group.17,18,22 Two studies showed a non-

significant trend that favored the early tracheostomy

group,20,21 while the remaining 3 suggested a slight increase

in pneumonia in the early tracheostomy group.15,16,19

Length of Mechanical Ventilation

Four studies reported the length of mechanical ventilation as an

outcome.17-19,23 The average length of mechanical ventilation

Table 1. Characteristics of Included Studies.

Author Year

Study

Population

No. of Patients in

Early Tracheosto

my Group

No. of Patients

in Late Tracheosto

my Group

Predictor of

Prolonged

Ventilation

Day of

Early

Tracheostomya

Day of

Late

Tracheostomyb

Tracheostomy

Method

Dunham and

LaMonica16

1984 Trauma 34 40 Clinical judgment 4 14 Open

Rodriguez

et al171990 Trauma 51 55 NS 7 8 NS

Sugerman

et al201997 Mixed 53 59 Clinical judgment 5 14 Both

Saffle et al19 2002 Burns 21 23 Validated

instrument

4 14 Open

Rumbak

et al182004 Medical 60 60 Clinical judgment 2 14 Percutaneous

Barquist

et al15

2006 Trauma 29 31 NS 8 28 Open

Terragni

et al212010 Mixed 209 210 Clinical

judgment

8 15 Percutaneous

Koch et al14 2012 Neuro, neurosurgery,

neurotr

50 50 Clinical

judgment

4 6 Percutaneous

Zheng

et al222012 Medical 58 61 NS 3 15 Percutaneous

Bosel

et al132013 Neurological

illness,

neurosurgery,

neurotrauma

29 31 Nonvalidated

instrument

3 7 Percutaneous

Young et al232013 Mixed 455 454 Clinical judgment 4 10 Both

Abbreviation: NS, not specified.aThe latest day on which a tracheostomy is performed in the early tracheostomy group.bThe earliest day on which a tracheostomy is performed in the late tracheostomy group.

Limits01/01/1983 09/01/2013Human researchEnglish language

Medline, Embase search terms:tracheo$ OR intubation AND time factors

Cochrane Central Register for Controlled Trialssearch terms:tracheostomy, tracheotomy

Title & Abstract Review

3873 studies excluded:Not pertaining to the timing oftracheostomy in patients requiringprolonged mechanical ventilationNot randomized or quasi-randomized trials in design

Full Text Review10 studies excluded:

Not prospective randomized orquasi-randomized trials in design

21 studies

516 duplicates removedMedline 2873 studiesEmbase 1226 studiesCENTRAL 310 studies

3893 studies

20 studies

11 studies to be included inreview

1 additional study found throughhand search of article bibliographies

Not comparing early and late

tracheostomy

Figure 1. Search strategy.

Liu et al 221







8/16/2019 liu et al 2015.pdf


among these studies was 17 days for the early tracheostomy

group and 24 days for the late tracheostomy group. A pooled

analysis was attempted, but once again, the heterogeneity was

found to be considerable ( I 2 = 99%); therefore, we present the

measured effects from the individual studies without a pooled

result (Figure 3). Rodriguez et al17 reported a significantly

decreased length of mechanical ventilation in the early tra-

cheostomy group, with a mean difference of 220 days (95%

CI, 220.8 to 219.2, P \ .05), as did Rumbak et al,18 with a

mean difference of 29.8 days (95% CI, 211.5 to 28.12,

Table 2. Risk of Bias of Studies Included in Meta-Analysis.

Author

Sequence

Generation

(Selection Bias)

Allocation

Concealment

(Selection Bias)

Blinding

(Performance

and Detection Bias)

Incomplete

Outcome Data

Addressed (Attrition Bias)

Selective

Outcome Reporting

(Reporting Bias) Other Bias

Dunham and

LaMonica16

High risk High risk Unclear risk High risk High risk High risk a

Rodriguez et al17 High risk High risk High risk High risk High risk Unclear risk b

Sugerman et al20 Low risk Low risk High risk High risk High risk High risk c

Saffle et al19 Low risk Low risk High risk Low risk High risk High risk d

Rumbak et al18 Low risk High risk Unclear risk Low risk Low risk Unclear risk b

Barquist et al15 Low risk Low risk Unclear risk Low risk Low risk High risk e

Terragni et al21 Low risk Low risk Low risk Low risk Low risk Unclear risk b

Koch et al14 Low risk Low risk Unclear risk Unclear risk Low risk Unclear risk b

Zheng et al22 Low risk Low risk High risk Low risk High risk High risk f

Bosel et al13 Low risk Low risk High risk Low risk High risk Low risk

Young et al23 Low risk Low risk High risk Low risk Low risk Low risk

aNo information provided on the baseline characteristics of the treatment groups.b

Where the time for follow-up is not provided for some outcomes, but it is unclear whether this would introduce bias to the study results.cSignificant amount of data collected but lost and therefore could not be analyzed.dEven after randomization, the 2 groups were not equal in all characteristics (eg, presence of full-thickness burns, probability of requiring prolonged ventilator

support).eIt appears that this study was terminated prior to completion as only the interim results are reported.f The patient flow diagram does not add up; it is unclear exactly how they arrived at 119 patients being randomized.

Figure 2. Incidence of pneumonia based on timing of tracheostomy. CI, confidence interval; M-H, Mantel-Haenszel.

Figure 3. Length of mechanical ventilation based on timing of tracheostomy. CI, confidence interval; IV, independent variable.







8/16/2019 liu et al 2015.pdf


P \ .001). Young et al23 did not find a significant

decrease in the length of mechanical ventilation, with a mean

difference of 21.6 days (95% CI, 23.33 to 0.13, P = .06).

Finally, Saffle et al19 found an increase in the length of

mechanical ventilation, with those in the early tracheostomy

group requiring 4.10 more days than those in the late tra-

cheostomy group (95% CI, 1.23 to 6.97, no P value reported).

ICU Length of Stay

Four studies reported data on ICU length of stay (Figure

4).17,18,20,23 The pooled results favored early tracheostomy, with a

weighted mean difference of 29.13 days (95% CI, 217.55 to

20.70, I 2 = 100%, P = .03). Sensitivity analysis showed that the

studies were equal in their contribution to the heterogeneity. The

study results were pooled despite the significant heterogeneity

because all of the measured effects were in the same direction,

favoring early tracheostomy; therefore, the true effect likely alsolies in this direction. A pooled analysis in this case likely captures

the true effect, despite significant statistical heterogeneity.

Hospital Mortality

To maximize the information captured and analyzed, we

defined mortality broadly in our protocol as hospital mortal-

ity. This definition included ICU, 28-day and 30-day mortal-

ity, and hospital mortality. Eight studies reported data on

mortality (Figure 5).15,17-23 The proportion of patients who

died in the early tracheostomy group was 30.7%, compared

with 34.7% in the late tracheostomy group. The weighted

absolute risk reduction was not significant. Similarly, there

was no significant reduction in the relative risk (RR, 0.84;

95% CI, 0.67 to 1.04; I 2 = 34%, P = .11). When sensitivity

analysis was performed for this outcome, the Rumbak et al18

study was found to contribute to most of the heterogeneity;

however, removing this study from the analysis did not

change the significance of the findings.

Sedation Use and Laryngotracheal Injury

Sedation use and laryngotracheal injury were also outcomes

of interest in our meta-analysis, as both are of clinical sig-

nificance. Data regarding these outcomes, however, could

only be qualitatively summarized as they were heteroge-

neous and could not be combined in a quantitative analysis.

Bosel et al13 found a significant decrease in the use of

sedatives in the early tracheostomy group compared with

the late tracheostomy group, with each group requiring

sedatives during 42% and 62% of the ICU stay, respectively

(median difference, 17.5 days; 95% CI, 3.3-29.2; P = .02).

Furthermore, they found that patients in the early tracheost-

omy group scored lower on the Richmond Agitation

Sedation Scale and spent more time under assisted rather

than controlled ventilation. Rumbak et al18 also found a sig-

nificant decrease in sedation use, with a mean (SD) of 3.2

(0.4) days of sedation in the early tracheostomy group com-

pared with a mean (SD) of 14.1 (2.9) days of sedation in the

late tracheostomy group ( P \ .001). Finally, Young et al23

found a significant decrease in the median number of days

of sedation use in the early tracheostomy group compared

with the late tracheostomy group (5 vs 8 days, P \ .001).

Figure 5. Hospital mortality based on timing of tracheostomy. CI, confidence interval; M-H, Mantel-Haenszel.

Figure 4. Intensive care unit (ICU) length of stay (LoS) based on timing of tracheostomy. CI, confidence interval; IV, independent variable.

Liu et al 223






8/16/2019 liu et al 2015.pdf


Only 3 studies investigated the presence and extent of lar-

yngotracheal injury as an outcome.16,18,20 The results from

these studies are presented in Table 3. Sugerman et al20 per-

formed laryngoscopy in 83 of 112 patients, either at the time

of tracheostomy performance or upon extubation. In those who

received a tracheostomy, 41 underwent laryngoscopy at the time

of decannulation. They found a nonsignificant trend toward ahigher incidence of airway ulceration and inflammation in the

late tracheostomy group. Specifically, they found a 35.3% inci-

dence of vocal cord ulceration in the early tracheostomy group

compared with 67.6% in the late tracheostomy group, while

subglottic inflammation was found in 0% and 39.3% of the

early and late tracheostomy groups, respectively. The incidence

of ulceration and inflammation at other laryngotracheal sites is

shown in Table 3. Patients who were symptomatic or had an

initial injury of Lindholm class II or worse were reassessed at 3

to 5 months. No evidence of late complications was found in

either the early or late tracheostomy group. Rumbak et al18 per-

formed airway assessments via bronchoscopy at the time of tra-cheostomy or extubation and at 10 weeks postintubation. There

was a trend toward higher rates of late tracheal stenosis in the

early tracheostomy group compared with the late tracheostomy

group (68.3% vs 38.3%); however, this was not statistically sig-

nificant. Finally, Dunham and LaMonica16 assessed for laryngo-

tracheal injury using flexible laryngoscopy following extubation

or decannulation. However, unlike the other 2 studies, they

assessed only patients who were symptomatic of upper airway

obstruction. They found early laryngotracheal injury in 17.6%

and 12.5% of the early and late tracheostomy groups, respec-

tively. This difference was not significant. Outcomes regarding

late laryngotracheal injury were not reported.

Subgroup Analysis

Subgroup analysis was performed to examine whether the

pooled results were influenced by the precise timing of the

tracheostomy and the illness population.

The analyzed studies varied in their definitions of early

tracheostomy, ranging from 2 to 8 days after intubation.

Table 4 summarizes the incidence of pneumonia, length of ventilation, length of ICU stay, and hospital mortality based

on the timing of the early tracheostomy. We did not find a

significant difference in any of the outcomes of interest

based on the timing of tracheostomy.

The studies encompassed a variety of patients, including

trauma, neurological, medical, and mixed medical, and sur-

gical. Therefore, subgroup analysis was performed with

regard to the etiology of critical illness to determine

whether specific populations of patients were more likely to

benefit from an early tracheostomy. A pooled analysis did

not find a significant difference in outcomes based on the

etiology of illness.

Discussion

Our meta-analysis suggests that when performed within 7

days, an early tracheostomy leads to a decreased length of

ICU stay compared with late tracheostomy. No difference was

seen in hospital mortality. A pooled analysis could not be per-

formed for the incidence of pneumonia and length of mechani-

cal ventilation secondary to significant heterogeneity. The

incidence of pneumonia as well as the length of mechanical

ventilation reported by the reviewed studies varied in direction,

with some reporting a benefit while others reported a detriment

Table 3. Incidence of Early and Late Laryngotracheal Complications Based on Timing of Tracheostomy (Expressed as Number of

Complications/Number of Evaluated Patients).

Early Complication Late Complication

Study Early Tracheostomy Late Tracheostomy Early Tracheostomy L ate Tracheostomy

Sugerman et al20a Epiglottic ulcers: 0/13

VC ulcers: 6/17Laryngeal ulcers: 0/13

Laryngeal inflammation: 26/33

SG ulcers: 1/13

SG inflammation: 0/15

Epiglottic ulcers: 8/32

VC ulcers: 23/34Laryngeal ulcers: 10/31

Laryngeal inflammation: 4/15

SG ulcers: 4/26

SG inflammation: 11/28

0 0

Rumbak et al18b TS

0%-20%: 52/60

21%-50%: 6/60

.50%: 3/60

TS

0%-20%: 41/60

21%-50%: 12/60

.50%: 5/60

TS

0%-20%: 26/60

21%-50%: 10/60

.50%: 5/60

TS

0%-20%: 13/60

21%-50%: 6/60

.50%: 4/60

Dunham and LaMonica16 LT path: 6/34 LT path: 5/40 NA NA

Abbreviations: LT path, laryngotracheal pathology, including subglottic stenosis, granuloma, and supraglottic edema; NA, not assessed in study; SG, subglottic;

TS, tracheal stenosis; VC, vocal cord.a

Late complications were assessed at 3 to 5 months postextubation in those who were symptomatic or who had an initial injury assessment of Lindholmclass II or worse. Sugerman et al20 presented varying denominators secondary to incomplete data for some outcomes. They also did not specify the number

of patients evaluated for late complications.bLate laryngotracheal complications assessed at 10 weeks postintubation.







8/16/2019 liu et al 2015.pdf


in the early tracheostomy group. Although the data regarding

sedation use did not lend themselves to a meta-analysis, all 3

studies examining this outcome found a significant decrease in

the early tracheostomy group.13,18,23

The risk of laryngotracheal injury is one of the chief con-

cerns for otolaryngologists when deciding to pursue to an

early tracheostomy. Our systematic review found that there

were no significant differences in the occurrence of early or

late injury based on the timing of tracheostomy.16,18,20

However, the included studies had small sample sizes. As

such, there is a high risk of type II error in these studies,whereby they were likely inadequately powered to detect

any true differences that might have existed between the

treatment groups. Our finding of no difference may reflect

the inadequacy of the current evidence rather than a true

absence of benefit in performing an early tracheostomy.

Another concern in deciding to perform an early tracheost-

omy is whether it actually leads to higher rates of long-term

complications, such as tracheal stenosis, especially in

patients who may not have needed to undergo the procedure

because of successful later extubation. The results from

Rumbak et al18 suggested a possible increase in late tracheal

stenosis in the early tracheostomy group, albeit this differ-ence was not significant. Therefore, at present, insufficient

evidence exists to substantiate or dispute this concern.

Studies with longer term follow-up are needed to assess the

true impact of an early tracheostomy on laryngotracheal out-

comes. This is an important issue that should be examined

in future investigations.

Our results differ from those of previous meta-analyses.

Griffith et al24 did not find a significant difference in mor-

tality or pneumonia, but they did find a significant decrease

in the duration of mechanical ventilation and ICU length of

stay. A subsequent meta-analysis by Wang et al25 did not

find a significant difference in any of the outcomes, includ-

ing mortality, pneumonia, length of mechanical ventilation,

and length of ICU stay. The varying results may be attrib-

uted to certain methodological differences. Our protocol

excluded studies that compared early tracheostomy with

prolon ged endotrach eal intubation, whereas both Griffith

et al and Wang et al included these studies in their overall

analysis. This can be problematic because in the studies

investigating early tracheostomy versus prolonged intubation,

data regarding the number of patients requiring a late tra-

cheostomy, the day of late tracheostomy, and these patients’

outcomes are often not explicitly stated. Therefore, it may be

inappropriate to combine the outcomes from these studies withthose looking at early versus late tracheostomy, as they are

measuring different treatment effects. Our meta-analysis also

included 4 additional trials that have been reported since the

Wang et al study. Finally, we employed a more conservative

and rigorous approach in the statistical analysis and interpreta-

tion of our data. Specifically, we chose to forego pooled analy-

ses for 2 of our outcomes due to significant heterogeneity. In

both the Griffith et al and Wang et al studies, data were com-

bined for every outcome even when significant heterogeneity

was found (with the I 2 statistic ranging from 58% to 87% in

the Griffith et al study and 0% to 98% in the Wang et al

study). The following paragraph discusses the issue of combin-ing data in the presence of high statistical heterogeneity.

A common limitation encountered in meta-analyses,

including the present one, is the heterogeneity of the data.

Heterogeneity can result from both clinical variability of the

study populations as well as methodological variability from

the performance of the studies.12 In meta-analyses examin-

ing the issue of early versus late tracheostomy, the major

sources of heterogeneity are the varying inclusion and

exclusion criteria employed by each trial, the diverse patient

populations, tracheostomy techniques, and particularly the

varied definitions of early and late tracheostomy. These var-

iations may result in more than one intervention effect. In

this case, a pooled result would not be appropriate, as it

Table 4. Outcomes Based on Timing of Early Tracheostomy.

Timing of Tracheostomy

Incidence of

Pneumonia

Length of Mechanical

Ventilation

Length of

ICU Stay

Hospital

Mortality

3 days18,22 RR = 0.38

95% CI = 0.13 to 1.16

I2 = 67%

P = .09

— — RR = 0.77

95% CI = 0.29 to 2.03

I2 = 71%

P = .59

4-5 days16,19,20,23 RR = 1.03

95% CI = 0.92 to 1.15

I2 = 0%

P = .58

WMD = 1.13

95% CI = 24.45 to 6.71

I2 = 91%

P = .69

WMD = 22.09

95% CI = 25.91 to 1.73

I2 = 96%

P = .28

RR = 0.97

95% CI = 0.83 to 1.13

I2 = 0%

P = .70

7-8 days15,17,21 RR = 0.87

95% CI = 0.64 to 1.19

I2 = 87%

P = .37

— — RR = 0.81

95% CI = 0.61 to 1.06

I2 = 0%

P = .13

Abbreviations: CI, confidence interval; I2, I-squared statistic as a measure of heterogeneity; ICU, intensive care unit; RR, relative risk; WMD, weighted mean

difference; —, insufficient data for pooled analysis.

Liu et al 225






8/16/2019 liu et al 2015.pdf


would not capture a real effect in any of the populations. In

our study, we decided to pool the data only if the heteroge-

neity was moderate or lower ( I 2 60%). The exception was

for ICU length of stay. We pooled the results for this out-

come as the measured effects from each study were in the

same direction, favoring early tracheostomy. In this case,

the true effect likely lies in that direction, and we felt that a

pooled analysis would best capture this.A second limitation of our study is that we may not have

included all randomized and quasi-randomized trials com-

paring the outcomes of early tracheostomy with late tra-

cheostomy. We performed a thorough search and minimized

the chance of missing relevant studies by having an inclu-

sive search strategy performed systematically by two

reviewers. The literature search, however, was limited to

articles in the English language; therefore, publications in

other languages may have been missed.

Finally, an important limitation of studies examining the

optimal timing of tracheostomies is the accuracy with which

the length of mechanical ventilation can be predicted. Moststudies included in our meta-analysis enrolled patients who

were predicted to require prolonged mechanical ventilation

based on the clinical acumen of the investigators. Two stud-

ies used institution-specific and nonvalidated tools. Only

one study used a validated tool; however, this tool is spe-

cific to burn patients and therefore is not applicable to other

critical care populations.26 Four studies provided data

regarding the number of late tracheostomies actually per-

formed.18,19,21,22 In these studies, a total of 354 patients

were randomized to receive a late tracheostomy, and of

these patients, 62 (17.5%) did not receive a tracheostomy as

they were successfully weaned from mechanical ventila-tion and extubated. Our inability to accurately predict the

need for mechanical ventilation is important to keep in

mind when considering early tracheostomies, as there are

a proportion of patients who may receive this intervention

unnecessarily. An important area for future research

would be the development of validated scoring systems,

allowing for more accurate prediction of prolonged

mechanical ventilation.

Conclusion

Our systematic review and meta-analysis found that when

performed within 7 days, an early tracheostomy was sig-nificantly associated with a decreased length of ICU stay.

There was no difference in hospital mortality. There is

not enough evidence at present to support an early tra-

cheostomy with regard to the incidence of pneumonia or

length of mechanical ventilation. Subgroup analysis did

not suggest a difference in outcomes based on the etiol-

ogy of critical illness or whether the early tracheostomy

was performed at 2 to 3 days, 4 to 5 days, or 7 to 8 days

of endotracheal intubation.

Acknowledgments

We thank Dr Brar for his assistance in devising the data collection

sheet and Dr Kassam for her help in reviewing our method.

Author Contributions

C. Carrie Liu, conception of study, acquisition of data, data analy-

sis, draft, revision and final approval of manuscript; Devon

Livingstone, acquisition of data, revision and final approval of

manuscript; Elijah Dixon, acquisition of data, revision and final

approval of manuscript; Joseph C. Dort, conception of study,

acquisition of data, revision and final approval of manuscript.Disclosures

Competing interests: None.

Sponsorships: None.

Funding source: None.

References

1. Pryor JP, Reilly PM, Shapiro MB. Surgical airway management

in the intensive care unit. Crit Care Clin. 2000;16:473-488.

2. Plummer AL, Gracey DR. Consensus conference on artificial

airways in patients receiving mechanical ventilation. Chest .

1989;96:178-180.3. Blot F, Similowski T, Trouillet JL, et al. Early tracheotomy

versus prolonged endotracheal intubation in unselected severely

ill ICU patients. Intensive Care Med . 2008;34:1779-1787.

4. Heffner JE, Hess D. Tracheostomy management in the chroni-

cally ventilated patient. Clin Chest Med . 2001;22:55-69.

5. Colice GL, Stukel TA, Dain B. Laryngeal complications of

prolonged intubation. Chest . 1989;96:877-884.

6. Whited RE. A prospective study of laryngotracheal sequelae in

long-term intubation. Laryngoscope. 1984;94:367-377.

7. McWhorter AJ. Tracheotomy: timing and techniques. Curr

Opin Otolaryngol Head Neck Surg . 2003;11:473-479.

8. Littlewood K, Durbin CG Jr.Evidenced-based airway manage-ment. Respir Care. 2001;46:1392-1405.

9. Juni P, Witschi A, Bloch R, Egger M. The hazards of scoring

the quality of clinical trials for meta-analysis. JAMA. 1999;

282:1054-1060.

10. Sharpe D. Of apples and oranges, file drawers and garbage:

why validity issues in meta-analysis will not go away. Clin

Psychol Rev. 1997;17:881-901.

11. Hartling L, Ospina M, Liang Y, et al. Risk of bias versus qual-

ity assessment of randomised controlled trials: cross sectional

study. Brit Med J . 2009;339:b4012.

12. Higgins JPT, Green S. Assessment of study quality. In: Cochrane

Handbook for Systematic Reviews of Interventions. Version 5.1.0.2011. http://www.cochrane-handbook.org. Accessed September 2,

2013.

13. 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. 2013;44:21-28.

14. Koch T, Hecker B, Hecker A, et al. Early tracheostomy

decreases ventilation time but has no impact on mortality of

intensive care patients: a randomized study. Langenbecks Arch

Surg . 2012;397:1001-1008.

15. Barquist ES, Amortegui J, Hallal A, et al. Tracheostomy in

ventilator dependent trauma patients: a prospective, rando-

mized intention-to-treat study. J Trauma. 2006;60:91-97.







8/16/2019 liu et al 2015.pdf


16. Dunham CM, LaMonica C. Prolonged tracheal intubation in

the trauma patient. J Trauma. 1984;24:120-124.

17. Rodriguez JL, Steinberg SM, Luchetti FA, Gibbons KJ, Taheri

PA, Flint LM. Early tracheostomy for primary airway manage-

ment in the surgical critical care setting. Surgery. 1990;108:

655-659.

18. Rumbak MJ, Newton M, Truncale T, Schwartz SW, Adams

JW, Hazard PB. A prospective, randomized, study comparing

early percutaneous dilational tracheotomy to prolonged trans-

laryngeal intubation (delayed tracheotomy) in critically ill

medical patients. Crit Care Med . 2004;32:1689-1694.

19. Saffle JR, SE Morris, Edelman L. Early tracheostomy does not

improve outcome in burn patients. J Burn Care Rehabil . 2002;

23:431-438.

20. Sugerman HJ, Wolfe L, Pasquale MD, et al. Multicenter, ran-

domized, prospective trial of early tracheostomy. J Trauma.

1997;43:741-747.

21. Terragni PP, Antonelli M, Fumagalli R, et al. Early vs late tra-

cheotomy for prevention of pneumonia in mechanically venti-

lated adult ICU patients: a randomized controlled trial. JAMA.2010;303:1483-1489.

22. Zheng Y, Sui F, Chen XK, et al. Early versus late percutaneous

dilational tracheostomy in critically ill patients anticipated requir-

ing prolonged mechanical ventilation. Chin Med J (Engl). 2012;

125:1925-1930.

23. Young D, Harrison DA, Cuthbertson BH, Rowan K. Effect of

early vs late tracheostomy placement on survival in patients

receiving mechanical ventilation: the TracMan randomized

trial. JAMA. 2013;309:2121-2129.

24. Griffith J, Barber VS, Morgan L, Young JD. Systematic

review and meta-analysis of studies of the timing of tracheost-

omy in adult patients undergoing artificial ventilation. BMJ .

2005;330:1243.

25. Wang F, Wu Y, Bo L, et al. The timing of tracheotomy in cri-

tically ill patients undergoing mechanical ventilation: a sys-

tematic review and meta-analysis of randomized controlled

trials. Chest . 2011;140:1456-1465.

26. Sellers BJ, Davis BL, Larkin PW, Morris SE, Saffle JR. Early

prediction of prolonged ventilator dependence in thermally

injured patients. J Trauma. 1997;43:899-903.

Liu et al 227






Documents

liu et al 2015.pdf