| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
* From the Departments of Medicine (Drs. Meade, Guyatt, Sinuff, and Cook) and Clinical Epidemiology & Biostatistics (Mss. Griffith and Hand), McMaster University, Hamilton, Ontario, Canada; and the Department of Respiratory Therapy (Ms. Toprani), Hamilton Health Sciences Corporation, Hamilton, Ontario, Canada.
Correspondence to: D.J. Cook, MD, McMaster University, Faculty of Health Sciences Center, Department of Clinical Epidemiology & Biostatistics, 1200 Main St West, Hamilton, Ontario, Canada; e-mail: debcook{at}mcmaster.ca
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Key Words: extubation • mechanical ventilation • meta-analysis • methods • modes • reintubation • systematic reviews • weaning
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Clinicians reach another stage in the weaning process at the point at which they suspect a patient may tolerate the discontinuation of mechanical ventilatory support and extubation. The most popular method of proceeding at this point is a T-piece trial. However, there are several alternatives, including supporting the patients through the trial of spontaneous breathing with low-level CPAP, PS, or positive end-expiratory pressure (PEEP). Within each of these approaches, numerous alternatives exist such as T-piece trials of various durations, and different levels of CPAP and PS.
In this systematic review, we begin with trials of the discontinuation of mechanical ventilation, continue with trials of stepwise reductions in mechanical ventilatory support, and, finally, present results of the third set of trials that focused on the last 48 h of the weaning process.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Search for Relevant Studies
To identify relevant studies, we searched MEDLINE, Excerpta
Medica Database, Health Services Technology Administration and
Research, Cumulative Index to Nursing and Allied Health Literature, the
Cochrane Controlled Trials Registry, and the Cochrane Data Base of
Systematic Reviews from 1971 to 1999. Search strategies are available
on request. We examined the reference lists of all included articles
for other potentially relevant citations. In addition, we hand searched
the journal Respiratory Care from 1997 to 1999. We did not
explicitly search for unpublished literature.
Two reviewers examined each title and abstract. We retrieved all articles that either reviewer considered to be possibly eligible. Two reviewers also examined the full text and made final decisions regarding eligibility based on the inclusion and exclusion criteria described earlier. These decisions were made unblinded to the source, authors, and conclusions of each study. The reviewers resolved disagreements by consensus.
Data Abstraction and Assessment of Methodological Quality
Two reviewers abstracted the data and assessed the
methodological quality of each study. Reviewers resolved disagreements
through discussion or consultation with one of the investigators. Final
data abstraction was rechecked by one of the investigators.
The methodological features of randomized trials that we abstracted included the following: the method of randomization and whether randomization was concealed; the criteria used to describe weaning, extubation, and reintubation; the extent to which groups were similar with respect to important prognostic factors; whether investigators conducted an intention-to-treat analysis; whether patients, clinicians, and those assessing outcome were blind to allocation; the extent to which the groups received similar cointerventions; and reporting of the reasons for study withdrawal.
Because it is a relatively new methodological term, the issue of concealment deserves further comment. A study is concealed if those making the judgment about whether a patient is eligible to be enrolled is unaware, at the time they are making this decision, whether a patient will be allocated to the experimental or control group. Lack of concealment theoretically may destroy the balance in prognostic factors that investigators strive to achieve through randomization. For instance, those making decisions about eligibility and enrollment may systematically exclude sicker patients if they know these patients are going to be allocated to the treatment group. Some empirical data suggest that unconcealed studies yield systematically larger treatment effects than do concealed studies.3 4
For nonrandomized controlled clinical trials, we considered the extent to which groups were similar with respect to important prognostic factors, whether the investigators adjusted for differences in prognostic factors, and the extent to which the groups received similar cointerventions.
Statistical Analysis
When we found duplicate reports of the same study in preliminary
abstracts and articles,5
6
7
8
9
10
we used the most complete
information.8
9
11
12
13
We divided our approaches to
synthesis (and subsequently our results) into studies dealing with
randomized and nonrandomized controlled comparisons of alternative
weaning interventions, and observational studies addressing the
prediction of successful weaning and the duration of mechanical
ventilation.
We began by identifying all the interventions and outcomes addressed by randomized trials. We abstracted or, when necessary, calculated effect sizes in terms of relative risks and associated 95% confidence intervals (CIs) for binary outcomes, and mean differences and 95% CIs for continuous variables. We transformed interquartile ranges into SDs when necessary to obtain variance estimates for testing differences between groups. The 95% CIs are not reported when variance data either were not provided by authors or were not estimable using either a precise p value or an interquartile range.
We reviewed the interventions and outcomes, and we decided when it was legitimate to pool across studies and when it was not. When pooling was not appropriate, we divided studies into categories according to the similarity of interventions.
We pooled data when, in our judgment, the underlying pathophysiology was such that across the range of populations, the management strategies in treatment and control groups, and the key outcomes studied we would expect more or less the same treatment effect. In general, we did not see differences in the distributions of characteristics of populations between studies as an impediment to pooling. For instance, consider two studies of the same weaning strategy. One enrolled patients a mean of 8 days after the onset of mechanical ventilation, another a mean of 14 days after onset. However, both studies enrolled some patients after 3 days of receiving mechanical ventilation, and others after 3 months. Thus, the individual studies are themselves pooling data from patients with the full spectrum of duration of mechanical ventilation. When interventions, outcomes, and trial methodology are similar, we do not see serious impediments to pooling in this sort of situation.
For instances in which we could pool data, for continuous variables we
considered the mean in each group and an estimate of variability from
each group which determined the weight given to the study in the pooled
analysis. For pooling binary data, we calculated risk ratios using the
methods described by Fleiss.14
We constructed 2 x 2
tables in each study for which the data were available and calculated
the associated risk ratios. For both continuous and binary variables,
we tested for heterogeneity using a test based on the
2 distribution with N -1 degrees of freedom,
in which N signifies the number of studies. When we found clinically
important heterogeneity that could not be explained by the play of
chance, we reviewed the methodology of the original studies in search
of explanations of the differences in outcome. All our analyses are
based on a random-effects model.
For nonrandomized studies that compared alternative weaning interventions, we used similar methodologies for calculating point estimates and 95% CIs for individual studies but made no attempt to pool data across studies.
|
Results |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In the second clinical context that we identified, the clinician faces a patient whose condition is improving, but who is unlikely to tolerate unassisted ventilation for at least 24 h, and perhaps up to several days. The clinician wishes to progressively decrease the level of mechanical support and has a number of modes available to achieve this goal. Five23 24 25 26 27 of the 16 randomized trials dealt with this context. We refer to this group of trials as "studies comparing alternative ventilation modes for stepwise reductions in mechanical support."
A third set of three trials focused on the last 48 h of the weaning process.28 29 30
Trials of Discontinuation Assessment Strategies
The sample size of the eight randomized studies of discontinuation
assessment strategy modes varied from 18 to 526. Three of the studies
enrolled > 100 patients (Table 1
).15
16
17
18
19
20
21
22
The largest studies21
22
were
methodologically strong, reporting the method of randomization and
weaning, extubation, and reintubation criteria. A lack of concealment
of randomization could easily bias the results of these studies, and,
while none of the studies used the optimal method of ensuring
concealment (ie, randomization by an independent methods
center), the studies by Esteban and coworkers21
22
used
the next best approach for a study such as this, opaque sealed
envelopes. Other randomized trials were not as strong methodologically
(Table 1)
.
|
|
|
The other randomized studies, all of which compared T-piece strategies to alternative strategies usually including some form of PS, had much smaller sample sizes, and generally had lower event rates. The result is very wide 95% CIs around any effect estimates (Table 2) . Our judgment was that we could pool only across two trials that compared T-piece breathing to CPAP, and, even after pooling, the number of events was so low that the 95% CIs were extremely wide (relative risk for nonextubation in CPAP vs T-piece breathing, 1.66 [95% CI, 0.60 to 4.64]; relative risk for reintubation, 1.61 [95% CI, 0.39 to 6.59]) (Table 3 ).
|
|
|
|
Of those randomized, patients in the trial by Esteban et al25 had a prior mean duration of mechanical ventilation of approximately 9.3 days, with a minimum of 24 h of ventilatory support. The patients in the study by Brochard et al23 also had a minimum duration of ventilation of 24 h, with an approximate mean duration of ventilation of 14 days.
Because the populations and interventions of the two studies are reasonably similar, we pooled the results for the outcomes that were measured in similar ways (Table 6 ). Table 6 presents with the duration of ventilation and the relative risk of the combined end point of nonextubation in 2 to 3 weeks and the need for reintubation. In the comparison of T-piece breathing to PS, the pooled results showed no difference in the duration of ventilation, the trends going in opposite directions in the two studies (Table 6) . The results of the trial by Esteban et al25 favored weaning with T-piece breathing, and those of the trial by Brochard et al23 favored PS. As a result, the 95% CI around the pooled estimates for both the duration of ventilation and the relative risk of nonextubation or reintubation is extremely wide.
|
In the comparison of PS to SIMV on the duration of weaning, both studies found trends in favor of PS, although the effect in the study by Brochard et al23 was much larger. The magnitude of the trend in the pooled result for the duration of ventilation was > 60 h, and the 95% CI came close to excluding no effect. The pooled results of PS vs SIMV on the combined end point showed an extremely wide 95% CI.
Jounieaux et al24 randomized 19 patients to SIMV with PS vs SIMV without PS. Neither group received CPAP. The duration of the weaning process was approximately 1 day shorter in the group that received PS, with the lower boundary of the 95% CI being approximately 7 h (Table 5) . Two patients in the SIMV group, and none in the group that also received PS, required reintubation.
Two groups of investigators26 27 evaluated NPPV as a mode for stepwise reductions in mechanical ventilatory support for patients admitted to the hospital with COPD exacerbations who had failed a 2-h T-piece breathing trial. The control strategies in the two studies included PS ventilation, with or without CPAP. In the larger study, Nava et al26 found that a reduction in the duration of mechanical ventilation, which was associated with a reduction in ICU stay of almost 9 days, was associated with NPPV (Table 5) . Pooling the results of these studies, the reduction in ICU length of stay was 5 days (95% CI, -12.2 to +1.9 days) (Table 7 ). Pooling also indicated favorable trends in mortality (relative risk, 0.30; 95% CI 0.09 to 1.02) and in the incidence of nosocomial pneumonia (relative risk, 0.29; 95% CI, 0.02 to 3.88).
|
Trials Comparing Alternative Ventilation Modes for Weaning Lasting
< 48 h
Three randomized trials addressed an intermediate group
of patients who were not yet ready for discontinuation assessment, but
who were likely to be ready within 48 h.28
29
30
These
trials were methodologically relatively weak and included sample sizes
of < 42 patients (Table 8
). Chopin et al28
showed a trend in favor of
CO2 mandatory ventilation over intermittent
mandatory ventilation (IMV) and multiple daily T-piece breathing in the
proportion of patients extubated at 24 h. However, there were only
14 patients in each of the three study groups (Table 9
).
|
|
The study by Esen et al30 showed trends in favor of PS over IMV in both the duration of ventilation and the success of extubation in 48 h (Table 9) .
Two non-RCTs33 34 also examined alternative weaning modes for patients who were expected to be weaned very quickly from mechanical ventilation, with similar findings to those of the RCTs. Tomlinson et al33 found no difference in the duration of mechanical ventilation or the duration of weaning in medical-surgical ICU patients who were weaned from ventilatory support over a period of 2 h to (rarely) 3 days using IMV (without CPAP) vs multiple daily T-piece breathing trials. In contrast, Rathgeber et al34 compared the following in the weaning of 586 patients who had undergone cardiac surgery: (1) the use of T-piece breathing trials; (2) the use of SIMV; or (3) the use of invasive bilevel pressure ventilation. The results of this study suggested a superiority of bilevel pressure ventilation (ie, CPAP plus PS) over weaning with T-piece breathing, and a superiority for both of these modes over weaning with SIMV, with respect to the duration of mechanical ventilation.
|
Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Controlled Trials of Alternative Discontinuation Strategies
Because of small sample sizes and low event rates, most of the
randomized trials have provided little information about the relative
impact of different modes of weaning. The results of the second largest
trial, by Esteban et al,21
have suggested a possible
advantage of PS over a T-piece trial of spontaneous breathing. The
results of the largest trial, also by Esteban et al,22
have suggested a possible advantage to 30-min T-piece breathing trials
over 2-h T-piece breathing trials with respect to ICU and hospital
lengths of stay.
A major theme that was found in other areas of weaning research emerges from review of the RCTs of unassisted ventilation in patients who the clinician anticipates are ready for extubation. Irrespective of the interventions being compared, studies examining the impact of short periods of spontaneous breathing with or without some form of support show very different rates of failure of extubation, or of reintubation. For instance, looking at the two largest trials, Esteban and colleagues21 found that 22% of 246 patients failed a T-piece weaning trial, and of the 192 who were extubated, 19% required reintubation. In contrast, Jones and coworkers17 reported that only 4% of 52 patients undergoing weaning with T-piece breathing were not extubated, and of those extubated, only 4% of 50 patients required reintubation.
These discrepancies suggest that investigators are using quite different criteria when judging whether a patient is ready for a trial of spontaneous breathing and for judging when the trial is a success and extubation is appropriate. When investigators explicitly described their criteria, the differences were not obvious. Nevertheless, differences in criteria must partially explain these differences in failure rates. Even if patients began with quite different unmeasured and undescribed differences in characteristics, clinicians must have been using different criteria for judging patient suitability for trials of unassisted breathing or extubation. Were this not the case, patients with a similar likelihood of succeeding would have been chosen for trials of unassisted breathing, or of extubation, from the initially dissimilar populations.
These varying criteria have a number of implications. First, because a particular weaning strategy is superior in a setting in which the threshold for a weaning trial, or for extubation, is low, does not mean that it will be superior in a setting in which the threshold is much higher. In fact, in those settings in which the threshold is high, and failure rates are < 5%, the absolute superiority of one approach over another will be small or perhaps negligible. The other implications of these varying criteria have to do with future research.
In the setting of a high threshold and low failure rates, investigators would need to recruit patient sample sizes in the thousands, or even the tens of thousands, to convincingly demonstrate differences between techniques. Such studies are unlikely to be feasible, and, even if they were feasible, they would consume substantial resources. Thus, investigators interested in studying the optimal use of ventilation strategies for weaning in the future should first establish plausible event rates, and if they are very low, should reconsider embarking on trials comparing alternative approaches. In situations in which event rates are high, it would be reasonable to reexamine whether a difference between PS and T-piece breathing could be confirmed.
Since the effect of different thresholds appears to overwhelm the impact of alternative ventilation approaches, perhaps investigators should study the use of these alternative thresholds. The randomized trials often specify that patients become eligible for studies when clinicians judge they are ready for decreasing ventilatory support or for a trial of unassisted ventilation. Since it is this decision that largely determines the outcome, the decision process or criteria warrant intense scrutiny. Some investigators have made their criteria explicit, but apparently similar explicit criteria yield very different results. Thus, there are subtleties in the decision-making process that have not yet been elucidated.
On the surface, it seems that clinicians who choose a high threshold and reduce failed trials of spontaneous breathing or reintubations to a minimum may be doing the best job. This is not necessarily so. Low rates of failure for breathing trials or reintubations are likely to come at the price of a prolonged duration of ventilation. Thus, a worthy goal for future research would be to delineate the tradeoff between prolongation of ventilation and failed trials of spontaneous breathing, and, more importantly, between prolongation of ventilation and reintubation. In other words, what is the cost, in terms of failed extubations, of decreasing weaning time by a particular amount?
Even if we knew the relationship between decreased weaning time and extubation failure, this would not in itself tell us the optimal tradeoff. For instance, were a study to find that a higher threshold decreased reintubation rates from 10 to 5% (we note that 570 patients per group would be required for an adequately powered study with these event rates) with patients spending a mean of an additional 24 h on the ventilator, which threshold would be judged to be optimal? Investigators interested in pursuing the issue of optimal ventilation strategies, and particularly an optimal threshold, should give careful thought to this issue.
Controlled Trials of Stepwise Reduction in Mechanical Ventilatory
Support
The issue of different thresholds that was so evident in the
randomized trials of unassisted ventilation is also relevant in
randomized trials of various modes for progressive reduction in
mechanical support. For instance, the mean duration of weaning in the
T-piece breathing group in the trial by Brochard et al23
was 8.5 days, and in the study by Esteban et al,25
3 days.
Herein, the major focus of judgment may be issues of patient selection
(although the reasons for differences are not apparent from the
descriptions of the recruiting process) and the judgment as to when the
weaning process begins.
The results of these studies suggest the possibility that weaning with multiple daily T-piece breathing, or PS, may be superior to SIMV. Even for this comparison, however, the 95% CIs on pooled estimates approach no effect. The results of two studies23 25 of weaning in < 48 h provide further evidence that SIMV may be less advantageous than other methods of decreasing mechanical ventilatory support. However, these trials compared particular SIMV weaning regimens. Other weaning regimens using SIMV may produce different results. This is also true of the other weaning modes studied. For instance, the criteria for weaning in the study by Esteban et al25 may have made it more difficult for patients in the PS group to meet extubation criteria than patients in the other groups.
The study by Jounieaux et al24 of SIMV and PS vs SIMV suggests the superiority of a regimen that includes PS by its finding of a shorter weaning time in the PS group. Because of the small sample size and low event rates, the study provides very little information about the effects on outcomes of nonextubation or reintubation.
The most dramatic finding in studies of progressive reduction in ventilatory support comes from two small, but sound, RCTs26 27 that suggest that NPPV is markedly superior to PS with or without CPAP. While these results are intriguing, the randomized trials that addressed this question enrolled a total of < 100 patients. However, the promising results of these studies are reinforced by the unequivocal results of systematic reviews of NPPV in patients with exacerbations of COPD who were on the threshold of ventilatory failure that demonstrate mortality benefits in favor of NPPV. An investigation of early NPPV in patients receiving mechanical ventilation should be one of the top priorities in this area of clinical research.
The data included in this systematic review and a more comprehensive discussion of the original articles are included in an Evidence Report of the Agency for Healthcare Research and Quality.35
|
Footnotes |
|---|
This article is based on work performed by the McMaster University Evidence-based Practice Center, under contract to the Agency for Healthcare Research and Quality (Contract No. 290-97-0017), Rockville, MD.
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  K. E.A. Burns, N. K.J. Adhikari, and M. O. Meade A meta-analysis of noninvasive weaning to facilitate liberation from mechanical ventilation: [Une meta-analyse d'un sevrage non effractif pour faciliter le retrait de la ventilation mecanique]. Can J Anesth, March 1, 2006; 53(3): 305 - 315. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. H. Petter, R. L. Chiolero, T. Cassina, P.-G. Chassot, X. M. Muller, and J.-P. Revelly Automatic "Respirator/Weaning" with Adaptive Support Ventilation: The Effect on Duration of Endotracheal Intubation and Patient Management Anesth. Analg., December 1, 2003; 97(6): 1743 - 1750. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
