Meta Analysis
Systematic review and meta-analysis of studies of the timing of tracheostomy in adult patients undergoing artificial ventilation.
This review may be edited
Posted By: Gordon S. Doig
Posted Date: 23 Aug 2005
Title: Systematic review and meta-analysis of studies of the timing of tracheostomy in adult patients undergoing artificial ventilation.
Authors: Griffiths J, Barber VS, Morgan L, Young JD.
Reference: BMJ. 2005 May 28;330(7502):1243. Epub 2005 May 18.
Link: Click here for a direct link to the paper. A password may be required for access to fulltext.
Abstract: OBJECTIVE: To compare outcomes in critically ill patients undergoing artificial ventilation who received a tracheostomy early or late in their treatment.
DATA SOURCES: The Cochrane Central Register of Clinical Trials, Medline, Embase, CINAHL, the National Research Register, the NHS Trusts Clinical Trials Register, the Medical Research Council UK database, the NHS Research and Development Health Technology Assessment Programme, the British Heart Foundation database, citation review of relevant primary and review articles, and expert informants.
STUDY SELECTION: Randomised and quasi-randomised controlled studies that compared early tracheostomy with either late tracheostomy or prolonged endotracheal intubation. From 15,950 articles screened, 12 were identified as "randomised or quasi-randomised" controlled trials, and five were included for data extraction.
DATA EXTRACTION: Five studies with 406 participants were analysed. Descriptive and outcome data were extracted. The main outcome measure was mortality in hospital. The incidence of hospital acquired pneumonia, length of stay in a critical care unit, and duration of artificial ventilation were also recorded. Random effects meta-analyses were performed.
RESULTS: Early tracheostomy did not significantly alter mortality (relative risk 0.79, 95% confidence interval 0.45 to 1.39). The risk of pneumonia was also unaltered by the timing of tracheostomy (0.90, 0.66 to 1.21). Early tracheostomy significantly reduced duration of artificial ventilation (weighted mean difference -8.5 days, 95% confidence interval -15.3 to -1.7) and length of stay in intensive care (-15.3 days, -24.6 to -6.1).
CONCLUSIONS: In critically ill adult patients who require prolonged mechanical ventilation, performing a tracheostomy at an earlier stage than is currently practised may shorten the duration of artificial ventilation and length of stay in intensive care.
Are the Results Valid?
1. Did the overview address a focused clinical question?
Yes. The authors express an interest in "randomised clinical trials involving the timing of trachosotomy in adult patients in intensive care units."
More specifically, they express an interest in trials of 'early' tracheostomy, defined as within 7 days of ICU admission, intubation or onset of ventilation.
2. Were the criteria used to select articles for inclusion appropriate?
They authors report "We selected studies for inclusion in the analysis if they were randomised or quasi-randomised clinical trials including adult patients requiring artificial ventilation."
It is well known that 'quasi-randomised' trials are likely to provide biased estimates of treatment effects by up to 40% ( Juni et al, BMJ 2001 Jul 7;323(7303):42-6.) It is likely that inclusion of these trials leads to significant bias.
3. Is it unlikely that important, relevant studies were missed?
The literature search was comprehensive and included Medline, EMBASE, other electronic resources and hand searching of the reference lists.
4. Was the validity of the included studies appraised?
No. Formal validity appraisal was not conducted. Results were not presented as 'high quality trials' vs all possible trials.
An extensive methodological review strongly suggests that validity of trials should be appraised on three key criteria: 1) the maintenance of allocation concealment, 2) the appropriate use of blinding and 3) the presentation of results in an intention to treat format. (Egger et al Health Technol Assess. 2003;7(1):1-76.)
Conclusions should be preferentially based on high quality trials.
5. Were assessments of studies reproducible?
Unclear. Measures of reproducibility were not presented.
6. Were the results similar from study to study?
The authors declare their threshold for the presence of 'significant heterogeneity' as an I2 measure of > 50%.
The authors report the presence of significant heterogeneity in all their meta-analyses (57.8%, 86.5%, 81.3%, 86.9%). At these levels of I2, it is likely that the overall pooled estimates of treatment effects are unreliable (Hatala et al CMAJ. 2005 Mar 1;172(5):661-5. ) even with the use of a random effects meta-analysis ( Villar et al Stat Med. 2001 Dec 15;20(23):3635-47.)
The source of this heterogeneity should be investigated. It is possible that the inclusion of pseudo-randomised trials, that do not maintain allocation concealment, explains the presence of this significant heterogeneity.
What are the Results?
1. What are the overall results of the overview?
Due to the presence of significant heterogeneity in all meta-analyses, it is likely that the pooled estimate of treatment effects are unreliable. We recommend conclusions be based on the review and critical appraisal of each individual trial.
The most reliable estimate of benefit/harm may be obtained by reviewing only the RCTs that actually maintained allocation concealment.
2. How precise were the results?
Will the Results Help Me In Caring For My Patients?
1. Can the results be applied to my patient care
Not in the current form.
2. Were all clinically important outcomes considered?
No. Post-extubation patient functional measures were not considered.
3. Are the benefits worth the harms and costs?

What other people had to say about:
Systematic review and meta-analysis of studies of the timing of tracheostomy in adult patients undergoing artificial ventilation.

Add Your Comment
[ Back ]

Any questions or comments please contact Gordon S. Doig
Implemented and designed by John Soer and Gordon Doig
Page last modified on Friday August 10, 2001