Original Research

Higher PEEP in Patients With Acute Lung Injury:

A Systematic Review and Meta-Analysis
Elliott C Dasenbrook MD MHS, Dale M Needham MD PhD,
Roy G Brower MD, and Eddy Fan MD

BACKGROUND: Studies of ventilation strategies that included higher PEEP in patients with acute
lung injury (ALI) or acute respiratory distress syndrome (ARDS) have yielded conflicting results.
OBJECTIVE: To determine whether higher PEEP during volume-limited and pressure-limited
ventilation is associated with 28-day mortality or barotrauma rates in patients with ALI/ARDS.
METHODS: We searched MEDLINE, CENTRAL, EMBASE, CINAHL, Web of Science, and the
bibliographies of retrieved papers to identify randomized controlled trials that compared higher
and lower PEEP in adult patients with ALI/ARDS who were already receiving volume-limited or
pressure-limited ventilation. Two of us independently abstracted study-level data, including study
design, patient characteristics, study methods, intervention, and main results. We pooled the study-
level data with a random-effects model, unless heterogeneity was low (I2 < 50%), in which case we
used a fixed-effects model. The primary outcome was 28-day mortality. RESULTS: Four random-
ized trials (2,360 participants) were evaluated. Higher PEEP had a nonsignificant trend toward
lower 28-day mortality (pooled relative risk 0.90, 95% CI 0.79 –1.02). There was no difference in
barotrauma between the 2 groups (pooled relative risk 1.17, 95% CI 0.90 –1.52). Two studies
reported an adjusted hospital death rate, and the pooled results of sensitivity analysis with those
adjusted rates were identical to those of the unadjusted analysis. CONCLUSIONS: In 4 recent
studies that used volume-limited or pressure-limited ventilation in ALI/ARDS patients, higher
PEEP was not associated with significantly different short-term mortality or barotrauma. This
study does not support the routine use of higher PEEP in patients with ALI/ARDS. Key words: acute
respiratory distress syndrome; adult; acute lung injury; mechanical ventilation; meta-analysis; mortality;
randomized controlled trial; review. [Respir Care 2011;56(5):568 –575. © 2011 Daedalus Enterprises]

Introduction

Acute lung injury (ALI) is a syndrome of life-threaten-

ing respiratory failure characterized by the acute onset of
Elliott C Dasenbrook MD MHS is affiliated with the Division of Pul- hypoxemia (PaO2/FIO2 ⱕ 300 mm Hg) and bilateral pul-
monary, Critical Care, and Sleep Medicine, Department of Medicine, monary infiltrates that are not primarily attributable to left
University Hospitals Case Medical Center, Case Western Reserve Uni- atrial hypertension.1 Acute respiratory distress syndrome
versity School of Medicine, Cleveland, Ohio. Dale M Needham MD
PhD, Roy G Brower MD, and Eddy Fan MD are affiliated with the
Division of Pulmonary and Critical Care Medicine, Department of Med-
icine, Johns Hopkins University School of Medicine, Baltimore, Mary-
SEE THE RELATED EDITORIAL ON PAGE 710
land. Dr Needham is also affiliated with the Department of Physical
Medicine and Rehabilitation, Johns Hopkins University School of Med-
icine, Baltimore, Maryland. (ARDS) is a subset of ALI with more severe hypoxemia
(PaO2/FIO2 ⱕ 200 mm Hg).1 ALI affects approximately
Dr Dasenbrook was partly supported by Cystic Fibrosis Foundation Clin- 190,000 patients each year in the United States, and the
ical Fellowship Research Award DASENB08D0. Dr Brower was partly
supported by National Institutes of Health grant P50 HL073994 and hospital mortality rate is approximately 39%.2
contract N01 HR56170. Dr Fan was partly supported by a fellowship Mechanical ventilation is essential for survival in most
award from the Canadian Institutes of Health Research. patients with ALI/ARDS. However, mechanical ventila-

568 RESPIRATORY CARE • MAY 2011 VOL 56 NO 5

 META-ANALYSIS OF HIGHER PEEP IN ACUTE LUNG INJURY

tion can also cause ventilator-induced lung injury (VILI), PEEP in adults with ALI/ARDS receiving volume-limited
which can delay or prevent recovery from acute respira- or pressure-limited ventilation, via meta-analysis of rele-
tory failure.3,4 One cause of VILI is excessive tidal volume vant randomized controlled trials that evaluated short-term
and pressure, which can overdistend aerated lung tissue.3-5 unadjusted mortality and barotrauma.
In a study by the National Institutes of Health ARDS
Network, a mechanical ventilation strategy with lower tidal Methods
volume and pressure was associated with a 9% absolute
lower short-term mortality in ALI patients, compared to a This study was performed at Johns Hopkins University
more traditional strategy that used larger tidal volume.6 School of Medicine, Baltimore, Maryland, and was con-
Another cause of VILI involves exhalation to a low lung ducted and is reported according to the Quality of Report-
volume and pressure,4,7 which injures small bronchioles ing of Meta-analyses (QUORUM) guidelines for meta-
and alveoli by repeated opening and closing during tidal analyses of randomized controlled trials.22
ventilation,7 and there may be excessive stress and strain
between aerated and atelectatic regions of lung paren- Data Sources and Search Strategy
chyma.8 The traditional approach to mechanical ventila-
tion involved modest PEEP (5–12 cm H2O) to prevent We electronically searched the MEDLINE, CENTRAL,
atelectasis and severe hypoxemia.9-11 However, some in- EMBASE, CINAHL, and Web of Science databases up to
vestigators recommend higher PEEP, to increase the pro- November 15, 2008, to identify potentially relevant pub-
portion of aerated lung at end-expiration (ie, maintain al- lications. Our search strategy included controlled vocabu-
veolar recruitment) and prevent VILI from exhalation to lary and related text words for: ALI/ARDS (study popu-
low volume and pressure.7,12 Moreover, higher PEEP may lation), use of PEEP (study intervention), and randomized
improve arterial oxygenation and allow a lower FIO2, which controlled trials (study design). The search strategy em-
could reduce pulmonary oxygen toxicity.13 However, these ployed standard filters for the identification of randomized
benefits of higher PEEP may be offset by additional lung clinical trials23,24 and included no language restrictions. In
injury due to overdistention or decreased cardiac output, addition, we hand-searched conference proceedings (2005
due to increased intrathoracic pressure and increased pul- through 2007) from the European Society of Intensive
monary vascular resistance.14 Care Medicine, American Thoracic Society, and the So-
The potential benefits of higher PEEP in patients with ciety of Critical Care Medicine, and the bibliographies of
ALI/ARDS already receiving volume-limited or pressure- all selected articles and relevant review articles to find
limited ventilation remain unclear, as existing randomized additional relevant abstracts and studies.
controlled trials may have been underpowered to find a
potentially small but clinically important reduction in short- Study Selection
term mortality.15-18 Two study-level meta-analyses have
been performed19,20: one concluded that higher PEEP was Eligible studies were randomized trials in which the
beneficial in unselected patients with ALI/ARDS.19 How- study groups received volume-limited or pressure-limited
ever, that study’s methods were suboptimal21 because those ventilation and either higher or lower PEEP in adult pa-
researchers pooled adjusted hospital mortality from one of tients (age ⱖ 18 y) with ALI or ARDS, as defined by, or
the studies15 but did not include the adjusted hospital mor- consistent with, the American-European Consensus Con-
tality data from another trial, in which there were also ference criteria.1 We excluded studies that only reported
imbalances in baseline characteristics.16 Our objective was physiologic and/or radiologic outcomes. Two reviewers
to evaluate the benefits and harms of higher versus lower (ECD and EF) independently screened titles, abstracts, and
studies for study eligibility, and disagreements were re-
solved via consensus. We assessed the reviewers’ agree-
Dr Dasenbrook presented a version of this paper at the International ment on study inclusion with the Cohen ␬ statistic.
Conference of the American Thoracic Society, held May 15-20, 2009, in
San Diego, California. Data Extraction
The authors have disclosed no conflicts of interest.
Two reviewers (ECD and EF) independently abstracted
Correspondence: Elliott C Dasenbrook MD MHS, Division of Pulmo- data and methods from the included studies, using stan-
nary, Critical Care, and Sleep Medicine, Department of Medicine, Uni- dardized forms. Abstracted data included study design,
versity Hospitals Case Medical Center, Case Western Reserve University patient characteristics, study methods, intervention, and
School of Medicine, 11100 Euclid Avenue, Mail Stop WRN 5067, Cleve-
land OH 44106. E-mail: ecd28@case.edu. main results. Differences in data abstraction were resolved
via consensus. The methodological quality of studies was
DOI: 10.4187/respcare.01011 evaluated according to published guidelines including: de-

RESPIRATORY CARE • MAY 2011 VOL 56 NO 5 569

 META-ANALYSIS OF HIGHER PEEP IN ACUTE LUNG INJURY

scription of randomization sequence generation; allocation

concealment; assessor blinding; completeness of outcome
data; and selective reporting of outcomes, eligibility cri-
teria, therapies, and excluded patients.25 Furthermore, study
quality was quantified with the Jadad score.26 Studies were
not excluded from the primary meta-analysis on the basis
of their quality assessment.
The primary outcome was all-cause mortality at 28 days.
Secondary outcomes were mortality in the intensive care
unit and hospital, and barotrauma (as defined in each trial).

Statistical Analysis

We report dichotomous outcomes as relative risk (RR)

and 95% confidence interval. If mortality at 28 days was
not explicitly stated, it was determined from data in the
published study.15 Studies with zero total events (ie, in
both the higher and lower-PEEP groups) were excluded
from the pooled analysis for that outcome.27,28 The I2 sta-
tistic was used as a measure of heterogeneity, calculated as
the proportion of total variation attributable to the be-
tween-study variation, and interpreted with published
guidelines: low heterogeneity 25– 49%, moderate hetero-
geneity 50 –74%, and high heterogeneity ⱖ 75%.29 A pri-
ori, we pooled study-level data with a random-effects Fig. 1. Study selection. ALI ⫽ acute lung injury. ARDS ⫽ acute
model,30 unless heterogeneity was low (I2 ⬍ 50%), in respiratory distress syndrome.
which case we used a fixed-effects model.31 We assessed
for publication bias with funnel plots and the Begg test.32 sure and pleural pressure, with pleural pressure estimated
A nominal P value of ⬍ .05 was taken as statistically from esophageal pressure).18
significant. The analyses were performed with statistics In the lower-PEEP group, PEEP was set in the follow-
software (Stata 10.0, StataCorp, College Station, Texas). ing ways: three used fixed combinations of PEEP and FIO2
to reach a target oxygenation goal,15,16,18 and one used the
Results lowest PEEP to target oxygenation and/or hemodynamic
goals.17 Plateau pressure on day 1 was greater in the high-
Search Results and Study Characteristics er-PEEP group (range 27–32 cm H2O) than in the lower-
PEEP group (21–25 cm H2O). Primary outcomes differed
The search identified 1,620 citations, and evaluation between the studies: the 28-day mortality range was 22–
found 4 eligible studies (Fig. 1).15-18 The 2 reviewers had 39% in the lower-PEEP groups, and 17–28% in the higher-
complete agreement (␬ ⫽ 1.0) on study inclusion. There PEEP groups (Table 2).
was no evidence of significant publication bias for the
primary or secondary outcome. The eligible studies were Quality Assessment
conducted in 5 different countries and enrolled a total of
2,360 (mean 590, range 61–983) adults with ALI/ARDS Overall, all the studies met most or all of the criteria for
(Table 1). Three of the studies were multicenter.15-17 All methodological quality (Table 3). All the studies met Amer-
the studies enrolled patients with both ALI and ARDS. ican-European Consensus Conference criteria for the di-
The studies used different mechanical ventilation strat- agnosis of ALI/ARDS,1 and had clearly defined eligibility
egies. In the higher-PEEP group, PEEP was set in the criteria, therapies, and reasons for patient exclusion. In all
following ways: two used tables of fixed combinations of the studies the investigators were not masked to treatment
PEEP and FIO2 settings to reach an oxygenation goal allocation after randomization.
range,15,16 and two set PEEP according to physiologic vari-
ables: one used the maximum values permitted while main- Evidence Synthesis
taining a plateau pressure ⬍ 30 cm H2O,17 and the other
adjusted PEEP according to end-expiratory transpulmo- None of the studies found a statistically significant dif-
nary pressure (ie, difference between airway-opening pres- ference in 28-day mortality between the PEEP groups.

570 RESPIRATORY CARE • MAY 2011 VOL 56 NO 5

 META-ANALYSIS OF HIGHER PEEP IN ACUTE LUNG INJURY

Table 1. Characteristics of the Included Randomized Controlled Trials

Baseline Target
Age Baseline PaO2/FIO2 Day 1 Day 1
PEEP APACHE II Target VT
First Author Year N mean ⫾ SD mean ⫾ SD PEEP Pplat
Arm Score PEEP (mL/kg
(y) (mm Hg) (cm H2O) (cm H2O)
mean ⫾ SD* PBW)

Brower15 2004 Higher 276 54 ⫾ 17 96 ⫾ 33† 151 ⫾ 67 Table‡ 6 14.7 27

Lower 273 49 ⫾ 17 91 ⫾ 30† 165 ⫾ 77 Table‡ 6 8.9 24
Meade16 2008 Higher 475 55 ⫾ 17 25 ⫾ 7 145 ⫾ 48 Table‡ 6 15.6 30
Lower 508 57 ⫾ 17 26 ⫾ 8 145 ⫾ 49 Table‡ 6 10.1 25
Mercat17 2008 Higher 385 60 ⫾ 16 ND 144 ⫾ 58 Pplat 28–30 6 14.6 28
Lower 382 60 ⫾ 15 ND 143 ⫾ 57 5–9 cm H2O§ 6 7.1 21
Talmor18 2008 Higher 30 55 ⫾ 16 26 ⫾ 6 147 ⫾ 56 Table㛳 6 18.7 32
Lower 31 51 ⫾ 23 27 ⫾ 7 145 ⫾ 57 Table‡ 6 11.0 25

* Higher Acute Physiology and Chronic Health Evaluation (APACHE) II score indicates greater severity of illness.
† APACHE III score.
‡ Table of fixed combinations of PEEP and FIO2.
§ Minimum PEEP guided by FIO2 and/or hemodynamics.
㛳 Table of fixed combinations of end-expiratory transpulmonary pressure and FIO2.
PBW ⫽ predicted body weight
Pplat ⫽ plateau pressure
ND ⫽ no data reported

Table 2. Outcomes of the Included Randomized Controlled Trials

28-Day Study Primary Outcome

First PEEP Days of Barotrauma
Year N Mortality Barotrauma Definition Primary Results (higher vs
Author Arm Follow-up no. (%)
no. (%) Outcome lower PEEP)

Brower15 2004 Higher 276 90 64 (23) Any new pneumothorax, 30 (11) Hospital 28% vs 25%
Lower 273 61 (22) pneumomediastinum, 27 (10) mortality P ⫽ .48
subcutaneous
emphysema, or
pneumatocele with a
diameter ⬎ 2 cm
Meade16 2008 Higher 475 75 135 (28) Pneumothorax, 53 (11) Hospital 36% vs 40%
Lower 508 164 (32) pneumomediastinum, 47 (9) mortality P ⫽ .19
pneumoperitoneum;
subcutaneous
emphysema on chest
radiograph; chest-tube
insertions for known
or suspected
spontaneous
pneumothorax
Mercat17 2008 Higher 385 60 107 (28) Pneumothorax between 26 (7) 28-day 28% vs 31%
Lower 382 119 (31) day 1 and 28 22 (6) mortality P ⫽ .31
Talmor18 2008 Higher 30 180 5 (17) Not defined 0 (0) PaO2/FIO2 at 280 vs 191
Lower 31 12 (39) 0 (0) 72 h P ⫽ .002

Mortality not associated with a significant difference in hospital mor-

tality (pooled RR 0.94, 95% CI 0.84 –1.05, P ⫽ .25,
The pooled analysis found a nonsignificant 28-day mor- I2 ⫽ 0%).15-17
tality trend that favored the higher-PEEP group (27%, Two studies adjusted for imbalances in baseline char-
n ⫽ 1,166 vs 30%, n ⫽ 1,194, pooled RR 0.90, 95% CI 0.79 – acteristics, including age and severity of illness.15,16 In the
1.02, I2 ⫽ 11%) (Fig. 2). Only one study reported inten- first study, adjustment for these baseline imbalances led to
sive care unit mortality, so a pooled analysis for that out- a nonsignificant trend toward lower in-hospital mortality
come was not possible.16 Hospital mortality was reported in the higher-PEEP group (25.1% vs 27.5%, P ⫽ .47).15 In
in 3 studies (n ⫽ 2,299 patients).15-17 Higher PEEP was the second study, adjustment for these imbalances, when

RESPIRATORY CARE • MAY 2011 VOL 56 NO 5 571

 META-ANALYSIS OF HIGHER PEEP IN ACUTE LUNG INJURY

Table 3. Methodological Quality of Included Studies*

Adequate Incomplete Free Free

Allocation Assessor Eligibility Excluded PEEP Jadad
First Sequence Outcome From From
Year Concealment Blinding Criteria Patients Therapy Quality
Author Generation Data Selective Other
Described Described Defined Described Described Score†
Described Addressed Reporting Bias

Brower15 2004 Yes Yes No Yes Yes Unclear‡ Yes Yes Yes 3
Meade16 2008 Yes Yes No Yes Yes Yes Yes Yes Yes 3
Mercat17 2008 Yes Yes No Yes Yes Yes Yes Yes Yes 3
Talmor18 2008 Yes Unclear§ No Yes Yes Unclear㛳 Yes Yes Yes 3

* Qualitative assessment was with Cochrane bias assessment method,25 in which each methods item is categorized as yes, no, or unclear.
† The Jadad quality score range is 1–5, and ⱖ 3 is considered high quality.26
‡ Stopped early for futility, imbalance in baseline characteristics between groups. Protocol modified after 171 enrollees.
§ Allocation concealment not described in primary manuscript.
㛳 Stopped early because of oxygenation benefit in intervention group.

Fig. 2. Association of higher PEEP and 28-day mortality. The size of the data marker indicates the weight of the study.

compared to the unadjusted results, increased the RR of volume-limited or pressure-limited ventilation, higher
in-hospital mortality in the higher-PEEP group, compared PEEP was not associated with lower short-term mortality
to the lower-PEEP group (RR 0.97, 95% CI 0.84 –1.12, or an increased risk of barotrauma.
P ⫽ .74).16 When we used the adjusted hospital mortality Our study differs from prior meta-analyses19,20,33 in sev-
from these studies in the analysis, the pooled results were eral ways. First, our study includes the most recently pub-
identical to the unadjusted analysis (pooled RR 0.94, lished randomized controlled trial.18 Second, for the pri-
95% CI 0.84 –1.05, P ⫽ .25, I2 ⫽ 0%).15-17 mary analysis we pooled unadjusted data. Third, we
performed a sensitivity analysis with adjusted data from
Barotrauma the 2 studies in which those data were available.15,16 In
contrast, 2 prior meta-analyses19,20 pooled adjusted hospi-
Barotrauma was reported in all of the studies, but was
tal mortality from one study15 (where adjusted mortality
excluded from the analysis in one study as there were no
favored higher PEEP), but did not include adjusted data
events in either group.18 Higher PEEP was associated with
from another trial that had imbalances in baseline charac-
a nonsignificant increase in barotrauma (pooled RR 1.17,
95% CI 0.90 –1.52, P ⫽ .25, I2 ⫽ 0%) (Fig. 3). These 4 teristics16 (adjusted data less favorable to higher PEEP).
studies included 109 (9%) barotraumas among 1,166 pa- These 2 meta-analyses consequently reported an effect of
tients in the higher-PEEP group and 96 (8%) barotraumas higher PEEP on hospital mortality as a pooled odds ratio
among 1,194 patients in the lower-PEEP group. of 0.86 (95% CI 0.72–1.02)20 and a pooled RR of 0.90
(95% CI 0.81–1.01).19 Despite a nonsignificant trend to-
Discussion ward benefit of higher PEEP, one of the 2 meta-analyses
concluded that the “current evidence supports the use of
In this meta-analysis of randomized trials of mechanical high PEEP in unselected groups of patients with ALI/
ventilation strategies in patients with ALI/ARDS receiving ARDS.”19 Our primary analysis, which pooled unadjusted

572 RESPIRATORY CARE • MAY 2011 VOL 56 NO 5

 META-ANALYSIS OF HIGHER PEEP IN ACUTE LUNG INJURY

Fig. 3. Association of higher PEEP and barotrauma. The size of the data marker indicates the weight of the study.

hospital mortality and reported an RR of 0.94 (95% CI 0.84 – patients with predominantly recruitable lung may benefit
1.05), resolves the conflicting conclusions from those 2 from a higher PEEP strategy, whereas those with predom-
prior meta-analyses, as our trend in unadjusted hospital inantly non-recruitable lung may not benefit and may be at
mortality cannot support the conclusion that unselected greater risk of VILI from overdistention.43 While none of
patients with ALI/ARDS may benefit from higher PEEP. the trials included in our study screened for recruitability
A recent meta-analysis that used patient-level data eval- at enrollment or during the treatment course, identification
uated the association between higher PEEP in adults with of PEEP responders may be useful for selecting a sub-
ALI and hospital mortality34 and reported an adjusted RR group of ALI/ARDS patients who might benefit from higher
of 0.94 (95% CI 0.86 –1.04) for hospital mortality with PEEP.39,44,45
higher PEEP.15-17 This result is very similar to our meta- Two studies reported important differences in patient
analysis result and reinforces the accuracy of our study- outcomes with higher PEEP: a significant increase in ven-
level meta-analysis, as compared to the other meta-analy- tilator-free and organ-failure-free days, and a nonsignifi-
ses that have used study-level data. Briel and colleagues cant mortality advantage,17 and a significant improvement
conclude that unselected patients with ALI/ARDS do not in 28-day mortality, after adjustment for baseline Acute
benefit from a higher-PEEP strategy, but the subgroup of Physiology and Chronic Health Evaluation II score
patients with severe hypoxemia (ie, ARDS patients) may (RR 0.46, 95% CI 0.19 –1.00, P ⫽ .049).18 These 2 studies
derive the greatest benefit from a higher PEEP strategy were similar in that they both used physiologic variables to
(adjusted RR for hospital mortality 0.90, 95% CI 0.81– adjust PEEP. One study increased PEEP until the plateau
1.00), and should be evaluated in future studies to confirm pressure was between 28 and 30 cm H2O,17 whereas the
if a mortality benefit is present.34 other used transpulmonary pressure to adjust PEEP.18 These
Experimental models7,35 and observational studies in hu- strategies probably delivered greater PEEP to responders
mans with ALI/ARDS36,37 suggest that higher PEEP can and lower PEEP to non-responders.46 In support of this
ameliorate VILI, but in the present systematic review we hypothesis, approximately 10% of patients actually had
found no significant differences in mortality with higher their PEEP lowered from the baseline value when guided
PEEP in ALI/ARDS patients. A potential explanation for by esophageal pressure.18 Importantly, titrating PEEP to
these discrepant results is the heterogeneous patient pop- oxygenation response (ie, with a table of fixed combina-
ulation captured by the current definition of ALI/ARDS.1 tions of PEEP and FIO2)15,16 may not lead to alveolar re-
Recent data suggest that there may be distinct subgroups cruitment, but to overdistention and an increase in VILI in
of ALI/ARDS patients with markedly different responses some patients.44 Despite 40 years of research, the optimal
to higher PEEP.38-42 Whole-body computed tomography level of and best approach for setting PEEP in ALI/ARDS
has demonstrated that higher PEEP in patients with a low patients remain elusive.46,47
percentage of recruitable lung (non-responders) provides Our study also suggests that there may be an increased
little benefit and may be harmful.39 In another study, non- risk of barotrauma with higher PEEP in ALI/ARDS pa-
responders (⬍ 150 mL alveolar recruitment) who received tients receiving volume-limited or pressure-limited venti-
a higher PEEP protocol similar to that used in one of the lation. Higher PEEP can increase plateau pressure, which,
studies in our systematic review15 experienced no change when greater than 35 cm H2O, is associated with a greater
in arterial oxygenation, but did experience significant in- risk of barotrauma.48 The fact that most patients in the 4
creases in static lung elastance.40 Therefore, ALI/ARDS included studies did not have plateau pressure higher than

RESPIRATORY CARE • MAY 2011 VOL 56 NO 5 573

 META-ANALYSIS OF HIGHER PEEP IN ACUTE LUNG INJURY

35 cm H2O may explain why there was no significant ACKNOWLEDGMENTS

difference in the risk of barotrauma. Another possibility is For methodological and statistical advice, we thank Steven N Good-
that, despite pooling, our study was underpowered to de- man MD MHS PhD, Department of Biostatistics, Elizabeth Colan-
tect a statistically significant difference in the barotrauma tuoni MSc PhD, Department of Biostatistics, Kay Dickersin PhD, De-
rate between the groups. For example, our meta-analysis partment of Epidemiology, and Tianjing Li MD, Department of
was powered (assuming a 2-sided alpha of .05 and power of Epidemiology, Johns Hopkins University, Baltimore, Maryland.
0.80) to detect a 3.6% absolute difference in barotrauma
REFERENCES
(11.6% vs 8.0%), but not the 1.3% difference in barotrauma
that we identified (9.3% vs 8.0%). Demonstrating statistical 1. Bernard GR, Artigas A, Brigham KL, Carlet J, Falke K, Hudson L,
significance for the identified 1.3% absolute difference in et al. The American-European Consensus Conference on ARDS.
Definitions, mechanisms, relevant outcomes, and clinical trial coor-
barotrauma between groups would require 14,878 patients.
dination. Am J Respir Crit Care Med 1994;149(3 Pt 1):818-824.
2. Rubenfeld GD, Caldwell E, Peabody E, Weaver J, Martin DP, Neff
M, et al. Incidence and outcomes of acute lung injury. N Engl J Med
Limitations
2005;353(16):1685-1693.
3. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from
experimental studies. Am J Respir Crit Care Med 1998;157(1):294-323.
First, our meta-analysis was underpowered to detect a small,
4. Tremblay LN, Slutsky AS. Ventilator-induced lung injury: from the
but potentially important, effect of higher PEEP on 28-day bench to the bedside. Intensive Care Med 2006;32(1):24-33.
mortality. For example, our meta-analysis was powered (as- 5. Fan E, Needham DM, Stewart TE. Ventilatory management of acute
suming a 2-sided alpha of .05 and power of 0.80) to detect a lung injury and acute respiratory distress syndrome. JAMA 2005;
5.2% absolute difference in 28-day mortality (24.6% vs 294(22):2889-2896.
6. Acute Respiratory Distress Syndrome Network. Ventilation with
29.8%), but not the 3.1% difference that we observed (26.7%
lower tidal volumes as compared with traditional tidal volumes for
vs 29.8%). Demonstrating statistical significance for the ob- acute lung injury and the acute respiratory distress syndrome. N Engl
served difference would require 6,566 patients. J Med 2000;342(18):1301-1308.
Second, many aspects of the intensive care in the studies 7. Muscedere JG, Mullen JB, Gan K, Slutsky AS. Tidal ventilation at
were not protocolized, with the exception of mechanical low airway pressures can augment lung injury. Am J Respir Crit
Care Med 1994;149(5):1327-1334.
ventilation. Since the investigators could not be blinded to
8. Mead J, Takishima T, Leith D. Stress distribution in lungs: a model
the PEEP treatment arm, differential treatment to patient of pulmonary elasticity. J Appl Physiol 1970;28(5):596-608.
groups could have resulted in important biases. 9. Carmichael LC, Dorinsky PM, Higgins SB, Bernard GR, Dupont
Third, barotrauma was variably defined and screened WD, Swindell B, et al. Diagnosis and therapy of acute respiratory
for in each trial, which may have been associated with distress syndrome in adults: an international survey. J Crit Care
1996;11(1):9-18.
misclassification and detection biases, respectively. Evi-
10. Esteban A, Anzueto A, Frutos F, Alia I, Brochard L, Stewart TE, et
dence for a detection bias may be suggested, as the inci- al. Characteristics and outcomes in adult patients receiving mechanical
dence of barotrauma was higher than that reported in ob- ventilation: a 28-day international study. JAMA 2002;287(3):345-355.
servational studies of ARDS patients.49 11. Thompson BT, Hayden D, Matthay MA, Brower R, Parsons PE.
Finally, moderate differences in age, baseline severity Clinicians’ approaches to mechanical ventilation in acute lung injury
and ARDS. Chest 2001;120(5):1622-1627.
of hypoxemia (ie, PaO2/FIO2), illness (ie, Acute Physiology
12. Webb HH, Tierney DF. Experimental pulmonary edema due to in-
and Chronic Health Evaluation score),50 and specific ven- termittent positive pressure ventilation with high inflation pressures.
tilation strategies may limit the generalizability of our find- Protection by positive end-expiratory pressure. Am Rev Respir Dis
ings to a particular population. 1974;110(5):556-565.
13. Altemeier WA, Sinclair SE. Hyperoxia in the intensive care unit: why
more is not always better. Curr Opin Crit Care 2007;13(1):73-78.
Conclusions 14. Vieillard-Baron A, Schmitt JM, Augarde R, Fellahi JL, Prin S, Page
B, et al. Acute cor pulmonale in acute respiratory distress syndrome
submitted to protective ventilation: incidence, clinical implications,
In adult patients with ALI/ARDS receiving volume-lim- and prognosis. Crit Care Med 2001;29(8):1551-1555.
ited or pressure-limited ventilation there was no signifi- 15. Brower RG, Lanken PN, MacIntyre N, Matthay MA, Morris A,
Ancukiewicz M, et al.; National Heart, Lung, and Blood Institute
cant association between higher PEEP and short-term mor-
ARDS Clinical Trials Network. Higher versus lower positive end-
tality or barotrauma. Future studies should investigate the expiratory pressures in patients with the acute respiratory distress
potential benefit of an approach designed to identify pa- syndrome. N Engl J Med 2004;351(4):327-336.
tients more likely to respond with recruitment, applying 16. Meade MO, Cook DJ, Guyatt GH, Slutsky AS, Arabi YM, Cooper DJ,
higher PEEP only to patients who are more likely to re- et al; Lung Open Ventilation Study Investigators. Ventilation strategy
using low tidal volumes, recruitment maneuvers, and high positive end-
spond with recruitment, and avoiding it in those less likely
expiratory pressure for acute lung injury and acute respiratory distress
to respond. Our study does not support the routine use of syndrome: a randomized controlled trial. JAMA 2008;299(6):637-645.
higher PEEP in patients with ALI/ARDS already receiving 17. Mercat A, Richard JC, Vielle B, Jaber S, Osman D, Diehl JL, et al;
volume-limited or pressure-limited ventilation. Expiratory Pressure (Express) Study Group. Positive end-expiratory pres-

574 RESPIRATORY CARE • MAY 2011 VOL 56 NO 5

 META-ANALYSIS OF HIGHER PEEP IN ACUTE LUNG INJURY

sure setting in adults with acute lung injury and acute respiratory distress acute lung injury and acute respiratory distress syndrome: systematic
syndrome: a randomized controlled trial. JAMA 2008;299(6):646-655. review and meta-analysis. JAMA 2010;303(9):865-873.
18. Talmor D, Sarge T, Malhotra A, O’Donnell CR, Ritz R, Lisbon A, 35. Slutsky AS. Lung injury caused by mechanical ventilation. Chest
et al. Mechanical ventilation guided by esophageal pressure in acute 1999;116(1 Suppl):9S-15S.
lung injury. N Engl J Med 2008;359(20):2095-2104. 36. Gattinoni L, D’Andrea L, Pelosi P, Vitale G, Pesenti A, Fumagalli R.
19. Phoenix SI, Paravastu S, Columb M, Vincent JL, Nirmalan M. Does Regional effects and mechanism of positive end-expiratory pressure
a higher positive end expiratory pressure decrease mortality in acute in early adult respiratory distress syndrome. JAMA 1993;269(16):
respiratory distress syndrome? A systematic review and meta-anal- 2122-2127.
ysis. Anesthesiology 2009;110(5):1098-1105. 37. Gattinoni L, Pelosi P, Crotti S, Valenza F. Effects of positive end-
20. Putensen C, Theuerkauf N, Zinserling J, Wrigge H, Pelosi P. Meta- expiratory pressure on regional distribution of tidal volume and re-
analysis: ventilation strategies and outcomes of the acute respiratory cruitment in adult respiratory distress syndrome. Am J Respir Crit
distress syndrome and acute lung injury. Ann Intern Med 2009; Care Med 1995;151(6):1807-1814.
151(8):566-576. 38. Ferguson ND, Kacmarek RM, Chiche JD, Singh JM, Hallett DC,
21. Dasenbrook EC, Fan E. Use of adjusted data in a meta-analysis Mehta S, et al. Screening of ARDS patients using standardized ven-
(letter). Ann Intern Med 2009;151(8):566-576. Erratum in: Ann In- tilator settings: influence on enrollment in a clinical trial. Intensive
tern Med 2009;151(12):897. http://www.annals.org/content/151/8/ Care Med 2004;30(6):1111-1116.
566/reply. Accessed March 8, 2011. 39. Gattinoni L, Caironi P, Cressoni M, Chiumello D, Ranieri VM,
22. Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF. Quintel M, et al. Lung recruitment in patients with the acute respi-
Improving the quality of reports of meta-analyses of randomised ratory distress syndrome. N Engl J Med 2006;354(17):1775-1786.
controlled trials: the QUOROM statement. Quality of Reporting of 40. Grasso S, Fanelli V, Cafarelli A, Anaclerio R, Amabile M, Ancona
Meta-analyses. Lancet 1999;354(9193):1896-1900. G et al. Effects of high versus low positive end-expiratory pressures
23. Wong SS, Wilczynski NL, Haynes RB. Developing optimal search in acute respiratory distress syndrome. Am J Respir Crit Care Med
strategies for detecting clinically sound treatment studies in EM- 2005;171(9):1002-1008.
BASE. J Med Libr Assoc 2006;94(1):41-47. 41. Villar J, Perez-Mendez L, Kacmarek RM. Current definitions of
24. Haynes RB, Wilczynski N, McKibbon KA, Walker CJ, Sinclair JC. acute lung injury and the acute respiratory distress syndrome do not
Developing optimal search strategies for detecting clinically sound reflect their true severity and outcome. Intensive Care Med 1999;
studies in MEDLINE. J Am Med Inform Assoc 1994;1(6):447-458. 25(9):930-935.
25. Higgins JPT, Altman DG. Chapter 8: Assessing risk of bias in in- 42. Villar J, Perez-Mendez L, Lopez J, Belda J, Blanco J, Saralegui I, et
cluded studies. In: Higgins JPT, Green S, editors. Cochrane hand- al. An early PEEP/FIO2 trial identifies different degrees of lung in-
book for systematic reviews of interventions, version 5.0.2 (updated jury in patients with acute respiratory distress syndrome. Am J Re-
September 2009). The Cochrane Collaboration; 2008. http://www. spir Crit Care Med 2007;176(8):795-804.
cochrane-handbook.org. Accessed March 8, 2011. 43. Slutsky AS, Hudson LD. PEEP or no PEEP: lung recruitment may be
26. Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, Gavaghan the solution. N Engl J Med 2006;354(17):1839-1841.
DJ, et al. Assessing the quality of reports of randomized clinical trials: 44. Grasso S, Stripoli T, De MM, Bruno F, Moschetta M, Angelelli G,
is blinding necessary? Control Clin Trials 1996;17(1):1-12. et al. ARDSNet ventilatory protocol and alveolar hyperinflation: role
27. Higgins JPT, Deeks JJ, Altman DG. Chapter 16: Special topics in of positive end-expiratory pressure. Am J Respir Crit Care Med
statistics. In: Higgins JPT, Greeen S, editors. Cochrane handbook for 2007;176(8):761-767.
systematic reviews of interventions, version 5.0.2 (updated Septem- 45. Rouby JJ, Puybasset L, Nieszkowska A, Lu Q. Acute respiratory
ber 2009). The Cochrane Collaboration; 2008. http://www.cochrane distress syndrome: lessons from computed tomography of the whole
handbook.org. Accessed March 8, 2011. lung. Crit Care Med 2003;31(4 Suppl):S285-S295.
28. Sweeting MJ, Sutton AJ, Lambert PC. What to add to nothing? Use 46. Gattinoni L, Caironi P. Refining ventilatory treatment for acute lung
and avoidance of continuity corrections in meta-analysis of sparse injury and acute respiratory distress syndrome. JAMA 2008;299(6):
data. Stat Med 2004;23(9):1351-1375. 691-693.
29. Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring in- 47. Gentile MA, Cheifetz IM. Optimal positive end-expiratory pressure:
consistency in meta-analyses. BMJ 2003;327(7414):557-560. the search for the Holy Grail continues. Crit Care Med 2004;32(12):
30. Dersimonian R, Laird N. Meta-analysis in clinical trials. Control 2553-2554.
Clin Trials 1986;7(3):177-188. 48. Boussarsar M, Thierry G, Jaber S, Roudot-Thoraval F, Lemaire F,
31. Mantel N, Haenszel W. Statistical aspects of the analysis of data from Brochard L. Relationship between ventilatory settings and barotrauma
retrospective studies of disease. J Natl Cancer Inst 1959;22(4):719-748. in the acute respiratory distress syndrome. Intensive Care Med 2002;
32. Begg CB, Mazumdar M. Operating characteristics of a rank corre- 28(4):406-413.
lation test for publication bias. Biometrics 1994;50(4):1088-1101. 49. Anzueto A, Frutos-Vivar F, Esteban A, Alía I, Brochard L, Stewart T,
33. Oba Y, Thameem DM, Zaza T. High levels of PEEP may improve et al. Incidence, risk factors and outcome of barotrauma in mechanically
survival in acute respiratory distress syndrome: a meta-analysis. Re- ventilated patients. Intensive Care Med 2004;30(4):612-619.
spir Med 2009;103(8):1174-1181. 50. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II:
34. Briel M, Meade M, Mercat A, Brower RG, Talmor D, Walter SD, et a severity of disease classification system. Crit Care Med
al. Higher vs lower positive end-expiratory pressure in patients with 1985;13(10):818-829.

This article is approved for Continuing Respiratory Care Education

credit. For information and to obtain your CRCE
(free to AARC members) visit
www.RCJournal.com

RESPIRATORY CARE • MAY 2011 VOL 56 NO 5 575