Lung protective ventilation strategy for the acute respiratory distress syndrome

  • Review
  • Intervention

Authors


Abstract

Background

Patients with acute respiratory distress syndrome and acute lung injury require mechanical ventilatory support. Acute respiratory distress syndrome and acute lung injury are further complicated by ventilator-induced lung injury. Lung protective ventilation strategies may lead to improved survival. This systematic review is an update of a Cochrane review originally published in 2003 and updated in 2007.

Objectives

To assess the effects of ventilation with lower tidal volume on morbidity and mortality in patients aged 16 years or older affected by acute respiratory distress syndrome and acute lung injury. A secondary objective was to determine whether the comparison between low and conventional tidal volume was different if a plateau airway pressure of greater than 30 to 35 cm H20 was used.

Search methods

In our previous 2007 updated review, we searched databases from inception until 2006. In this third updated review, we searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, CINAHL and the Web of Science from 2006 to September 2012. We also updated our search of databases of ongoing research and of reference lists from 2006 to September 2012.

Selection criteria

We included randomized controlled trials comparing ventilation using either a lower tidal volume (Vt) or low airway driving pressure (plateau pressure 30 cm H2O or less), resulting in a tidal volume of 7 ml/kg or less, versus ventilation that used Vt in the range of 10 to 15 ml/kg in adults (16 years old or older) with acute respiratory distress syndrome and acute lung injury.

Data collection and analysis

We independently assessed trial quality and extracted data. Wherever appropriate, results were pooled. We applied fixed-effect and random-effects models.

Main results

We did not find any new study which were eligible for inclusion in this update. The total number of studies remained unchanged, six trials involving 1297 patients. Five trials had a low risk of bias. One trial had an unclear risk of bias. Mortality at day 28 was significantly reduced by lung-protective ventilation with a relative risk (RR) of 0.74 (95% confidence interval (CI) 0.61 to 0.88); hospital mortality was reduced with a RR of 0.80 (95% CI 0.69 to 0.92). Overall mortality was not significantly different if a plateau pressure less than or equal to 31 cm H2O in the control group was used (RR 1.13, 95% CI 0.88 to 1.45). There was insufficient evidence for morbidity and long-term outcomes.

Authors' conclusions

Clinical heterogeneity, such as different lengths of follow up and higher plateau pressure in control arms in two trials, makes the interpretation of the combined results difficult. Mortality was significantly reduced at day 28 and at the end of the hospital stay. The effects on long-term mortality are unknown, although the possibility of a clinically relevant benefit cannot be excluded. Ventilation with lower tidal volumes is becoming a routine strategy of treatment of acute respiratory distress syndrome and acute lung injury, stopping investigators from carrying out additional trials.

Plain language summary

A gentler form of mechanical breathing for people affected by severe lung failure

Critically ill people affected by severe, acute respiratory failure need air to be pumped into their lungs (mechanical ventilation) to survive. Mechanical support buys time for the lungs to heal. Nevertheless, 35% to 65% still die. Several studies have suggested that mechanical breathing can also cause lung damage and bleeding. A new lung protective way of mechanical ventilation was tested in large studies. In this third update of the Cochrane review we searched the databases until September 2012 but we did not find any new study which was eligible for inclusion. The total number of studies remained unchanged, six trials involving 1297 people. This systematic review shows that a gentler form of mechanical breathing (so-called protective ventilation) can decrease deaths in the short term, by 26% on average, but the effects in the long term are uncertain or unknown.

Background

Description of the condition

Acute respiratory distress syndrome (ARDS) is a common, devastating clinical syndrome of acute lung injury that affects both medical and surgical patients. It is clinically characterized by severe hypoxaemia, radiographic evidence of bilateral pulmonary infiltration and absence of left heart failure. Acute lung injury (ALI) is a subset of ARDS with less severe impairment in oxygenation. A recent study has estimated the incidence of this disease to be between 15 and 34 cases per 100,000 inhabitants per year (Frutos-Vivar 2004). According to a prospective cohort study, the prevalence of ARDS and ALI was about 9% amongst intensive care patients, and 39.6% amongst ventilated patients (Roupie 1999). The mortality rate from ARDS and ALI is approximately 35% to 65% (Esteban 2008). There is also short- and long-term morbidity associated with these syndromes. Short-term morbidity leads to a prolonged stay in the intensive care unit and prolonged ventilator dependence (Davidson 1999). Those who survive the illness have a reduced health-related quality of life as well as cognitive impairment and high rates of disability (Dowdy 2006).

Recently, a new definition of ARDS, based on the degree of hypoxaemia, was proposed (Ranieri 2012).

Description of the intervention

Mechanical ventilation (MV) represents the main therapeutic support to maintain acceptable pulmonary gas exchange whilst treating the underlying disease. ARDS is associated with a reduction in static compliance (Marini 1990). As a result of this low compliance, high pressures are needed to obtain a sufficient tidal volume. The larger the tidal volume the higher the pressure required, which may lead to barotrauma, that is, alveolar rupture and radiological evidence of extra-alveolar air. In such patients, mechanical ventilation could lead to injury due to over-distension. This results from the distribution of the increased tidal volume to the high-compliance regions causing stretching and sheer forces on the alveolar wall (volutrauma) (Dreyfuss 1998). Plateau pressure, defined as airway pressure during the end-expiratory pause, roughly reflects the level of alveolar over-distension. Finally, parts of the lung are consolidated and are thus not recruitable. This leads to fewer alveoli truly being ventilated; hence a 'normal' tidal volume may not be appropriate for those remaining alveoli. Damage caused by MV to the lungs has been termed ventilator-induced lung injury (VILI).

How the intervention might work

Better understanding of the pathophysiology of ARDS and of VILI have led to the proposal that airway pressure and tidal volume should be limited in managing the ventilation of ARDS patients (Artigas 1998; Hickling 1990; Hickling 1994). This entails accepting a rise in the arterial partial pressure of carbon dioxide. In addition, cyclic inflation-deflation of injured lung units or alveoli can exacerbate lung injury (Dreyfuss 1998), and medium to high levels of positive end-expiration pressure (PEEP) should be used to keep alveoli open throughout the ventilatory cycle. Overall, this type of approach has been termed a lung protective ventilation strategy (Artigas 1998). Ventilation with lower tidal volumes was also associated with lower levels of systemic inflammatory mediators (Ranieri 1999).

Why it is important to do this review

Lowering the tidal volume is possibly not without hazards (Roupie 1999). Severe hypercapnia and acidosis can have adverse effects, including increased intracranial pressure, depressed myocardial contractility, pulmonary hypertension and depressed renal blood flow (Feihl 1994). The view that these risks are preferable to the higher plateau pressure required to achieve normocapnia has represented a substantial shift in ventilatory management.

Objectives

The objective of this systematic review was to compare ventilation with a lower tidal volume, with lower or higher PEEP, and ventilation with conventional tidal volume, with lower or higher PEEP, to determine whether such a lung protective ventilation strategy (LPVS) reduced morbidity and mortality in critically ill adults affected by ALI or ARDS.

Methods

Criteria for considering studies for this review

Types of studies

We accepted only randomized controlled trials (RCTs) to guarantee control of selection bias. We excluded studies that, on closer scrutiny, were determined to be quasi-randomized or cross-over studies.

Types of participants

We included critically ill patients aged 16 years or older, intubated and ventilated and affected by ARDS or ALI from any cause as defined by the North-American-European Consensus Conference on ARDS (NAECC) (Bernard 1994) or by the Lung Injury Severity Score (LISS) (Murray 1988).

Types of interventions

A protective ventilation strategy that used lower tidal volume or low airway driving pressure (plateau 30 cm H2O or less), or a combination of the two, which resulted in a tidal volume of 7 ml/kg or less versus conventional mechanical ventilation that used a tidal volume in the range of 10 to 15 ml/kg. Protective ventilation or conventional ventilation may include lower or higher levels of PEEP. Regardless of the strategy used to deliver the interventions, the two study groups had to differ only in the tidal volume and not for other elements of the associated ventilatory strategy. Hypercapnia (as an unavoidable part of the protective ventilation intervention) was accepted as long as the resulting acidosis was controlled and kept within acceptable ranges and those ranges were clearly stated.

Types of outcome measures

Primary outcomes

Overall mortality, evaluated at hospital discharge (if this information was unavailable, mortality was evaluated at the end of the follow-up period scheduled for each trial)

There is evidence that the cause of death in ARDS is the development of organ failure (Monchi 1998), and that mechanical ventilation might be a factor leading to multiple organ failure. The duration of mechanical ventilation could also lead to the development of associated infectious diseases that could affect the patients' overall prognosis, such as ventilator-associated pneumonia (VAP). If mortality is equal, other outcomes become important to the patients such as long-term quality of life and cognitive impairment.

Secondary outcomes

1. Development of multi-organ failure (MOF)
2. Duration of mechanical ventilation and total duration of mechanical support
3. Total duration of stay in intensive care unit and hospital
4. Long-term mortality
5. Long-term health-related quality of life
6. Long-term cognitive outcome

Search methods for identification of studies

Electronic searches

In our original review (Petrucci 2004a) we searched the databases from inception until 2006.

In this updated review we searched the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library 2012, Issue 3 (Appendix 1); MEDLINE (2006 to September 2012) (Appendix 2); EMBASE (2006 to September 2012) (Appendix 3); CINAHL (2006 to September 2012) (Appendix 4) and the Web of Science (2006 to September 2012) (Appendix 5) using a combination of MeSH and text words.

We applied no language restrictions.

In addition, we handsearched references lists and abstracts and proceedings of scientific meetings held on the subject.

In our original review (Petrucci 2004a), we searched proceedings of the Annual Congress of the European Society of Intensive Care Medicine (ESICM) and of the American Thoracic Society (ATS) from 1993 to 2006. In this version we updated that search to September 2012.

Searching other resources

We searched the following web resources:

We contacted the original author(s) for clarification about content, study design and missing data, if needed.

Data collection and analysis

Selection of studies

Nicola Petrucci (NP) and Carlo De Feo (CDF) independently screened the results of the search strategy for potentially relevant trials and independently assessed them for inclusion based on the inclusion criteria. We resolved disagreements through discussion until a consensus was reached.

Data extraction and management

We employed the standard methods of the Cochrane Anaesthesia Review Group. We (NP, CDF) independently performed assessment of methodology and extraction of data, with comparison and resolution of any differences found at each stage.

Assessment of risk of bias in included studies

We judged the quality of each trial by whether or not the study design had minimized bias within the scope of the clinical context. The risk of bias in included studies was assessed by addressing six specific domains: random sequence generation, allocation concealment, blinding, incomplete outcome data, selective reporting, and other issues.

We defined high-quality trials as: controlled, appropriately randomized, having adequate concealment of allocation, and completeness of follow up according to an intention-to-treat analysis. We judged the concealment of allocation as adequate if the trials took adequate measures to conceal allocation through central randomization, such as serially numbered opaque envelopes or a table of random numbers. We judged the generation of allocation sequence as adequate where trials were deemed to have a satisfactory sequence generation (random numbers generated by computers, drawing of lots of envelopes).

For the individual trial results, we calculated relative risk (RR) with 95% confidence interval (CI) and number needed to treat (NNT) for categorical outcomes. We reported mean differences (MDs) (and 95% CIs) for continuous variables, if appropriate.

Measures of treatment effect

For the meta-analysis, we reported relative risk (RR), risk difference (RD) and 95% CI for dichotomous outcomes; and weighted mean differences (WMD) for continuous variables.

Unit of analysis issues

We did not have any unit of analysis issues.

Dealing with missing data

For all included trials, we registered the levels of attrition. We tried to contact main authors for missing data.

Assessment of heterogeneity

We performed a formal exploration of heterogeneity. Higgins developed a measure of the impact of heterogeneity in a meta-analysis that is independent of the number of studies (I2 statistic) (Higgins 2002). This measure describes the proportion of total variation in study estimates that is due to heterogeneity rather than sampling error. We used the I2 statistic as an interpretation of the extent of heterogeneity. Values of I2 less than 56% and higher than 31% account for 'moderate' heterogeneity. The possibility of a type II (false negative) error must be considered and a thorough attempt was made to identify clinical heterogeneity or sources of bias.

Assessment of reporting biases

We explored publication bias using a plot of study sample size against RR. In the absence of publication bias, a plot of study sample size (or study weight) versus outcome (that is log RR) should have a bell or inverted funnel shape with the apex near the summary effect estimate (funnel plot).

Data synthesis

We pooled the results from the original RCTs using a fixed-effect model, or a random-effects model for I2 between 31% and 56% (Deeks 2001).

Subgroup analysis and investigation of heterogeneity

We planned subgroup analysis to determine whether the results differed by one of the following.

Population
  1. Age

  2. Severity of disease: ARDS (more severe impairment) or ALI (less severe impairment)

  3. Aetiology of ARDS and ALI (i.e. pneumonia, trauma, sepsis, etc)

Delivery of interventions
  1. To determine whether the comparison between low tidal volume and normal tidal volume was different if a plateau pressure of greater than 30 to 35 cm H20 was used

We performed a subgroup analysis based on the plateau pressure in the control groups.

Sensitivity analysis

We also performed sensitivity analyses based on whether the outcome assessors were blinded and by imputing values for dropouts.

Results

Description of studies

Results of the search

In our previous updated review (Petrucci 2007) we found 10 studies that were of potential relevance, of which we eventually excluded four studies (Brower 2004; Esteban 2000; Ranieri 1999; Rappaport 1994). In this updated review we searched the literature from 2007 to 2012 (Figure 1).

Figure 1.

Study flow diagram.

Included studies

Six studies in total met our study inclusion criteria (Amato 1998; ARDS Network 2000; Brochard 1998; Brower 1999; Stewart 1998; Villar 2006) and were assessed for methodological quality.

We did not find any studies, in addition to our previous six trials (Amato 1998; ARDS Network 2000; Brochard 1998; Brower 1999; Stewart 1998; Villar 2006), that were assessed for methodological quality (see Characteristics of included studies table). All of the included studies were multi-centre trials.

The total number of patients randomized in each study varied from 52 (Brower 1999) to 861 (ARDS Network 2000). All studies included ARDS patients but some investigations tended to use the LISS definition (Amato 1998; Brochard 1998) and some the NAECC definition of ARDS (ARDS Network 2000; Brower 1999; Villar 2006). The average age at randomization with standard deviation (SD) varied from 33 ± 13 (Amato 1998) to 59 ± 17 (Stewart 1998). The European study (Brochard 1998) was unique in the inclusion of patients with single organ failure (lung injury) only. The time that had elapsed from eligibility to randomization ranged from one hour to 36 hours. In one study (Villar 2006), only patients who demonstrated persistent ARDS 24 hours after initially meeting the ARDS criteria were enrolled.

All the trials assessed the baseline risk by combining several prognostic variables into a severity score (APACHE II or APACHE III score), which provides initial risk stratification for severely ill, hospitalized patients and a risk estimate for hospital mortality for individual intensive care unit (ICU) patients (Knaus 1985; Knaus 1991). The higher the score the higher is the relative risk of hospital death. The APACHE II score at baseline was in the range of 17 ± 8 SD (Brochard 1998) to 28 ± 7 SD (Amato 1998). Two trials used the APACHE III score (ARDS Network 2000; Brower 1999). The APACHE III score ranged from 81 ± 28 SD (ARDS Network 2000) to 90 ± 26 SD (Brower 1999). The severity of impairment of lung function was reported according to the partial pressure of arterial oxygen to the fraction of inspired oxygen ratio (PaO2:FiO2) (see 'Characteristics of included studies' table). Patients with less severe hypoxaemia (as defined by a PaO2: FiO2 of 300 or less) are considered to have ALI, and those with more severe hypoxaemia (as defined by a ratio of 200 or less) are considered to have ARDS (ARDS Conference 1977).

Delivery of the interventions for each trial varied. Five trials (ARDS Network 2000; Brochard 1998; Brower 1999; Stewart 1998; Villar 2006) set the tidal volume based on body weight, without the use of a pressure-volume (PV) curve. However, in the ARDS Network, Brower, Stewart and Villar trials the tidal volume was set according to predicted (or ideal) body weight (IBW), which was calculated according to the gender and height of the patient. IBW is, on average, 20% less than measured body weight. When transformed to ml/kg of measured body weight the mean tidal volume in the ARDS Network trial ranged from 9.4 to 9.9 ml/kg in the control group, which was quite similar to values used in the other trials, and 5.2 ml/kg in the low tidal volume group, which was lower than in other trials. Two trials (Amato 1998; Villar 2006) compared the effect of a combined strategy composed of a low tidal volume and relatively high PEEP titrated according to the PV curve. Amato added an intermittent recruitment manoeuvre, and used in the control group a ventilatory strategy aimed at normalizing the partial pressure of carbon dioxide, without any limitation in peak inspiratory pressure. Based on aggregate data, all patients in the lung protective ventilation groups received tidal volumes significantly lower than those in the conventional ventilation groups. Limits of airway plateau pressure ranged from 22 to 30 cm H2O in the protective arms and from 31 to 37 cm H2O in the conventional arms. Most studies focused on plateau airway pressure, whereas one study focused on peak inspiratory airway pressure (Brochard 1998). Protocols for the management of acidosis using bicarbonate infusions were developed in all trials, although there were some differences. The ARDS Network investigators (ARDS Network 2000) were most aggressive in attempting to keep the pH greater than 7.30 for all patients and allowed violations of the tidal volume and airway pressure limits when the pH fell below 7.15. In contrast, another study (Stewart 1998) did not dictate volume and pressure violations until the pH fell to 7.00. Co-interventions were not detailed in the studies. Only one trial (Amato 1998) reported treatments other than ventilation. None of the studies reported use of prone positioning or surfactant. Nitric oxide was used in one trial only (Brochard 1998) and did not appear to be an important feature that imbalanced the trial as it was used in about one-fifth of patients in both groups.

The studies did not report all the considered outcomes. The definition of overall mortality differed between studies. Mortality was measured at a cut-off point of day 60 in one study (Brochard 1998), day 180 in one study (ARDS Network 2000) and at hospital discharge in two studies (Brower 1999; Stewart 1998). In one study (Amato 1998) both mortality at day 28 and at hospital discharge were reported. ICU and hospital mortality were reported in one study (Villar 2006). Importantly, in ARDS Network 2000 patients were followed until discharged home or for 180 days, whichever occurred first. This may be considered equivalent to hospital mortality. Among the secondary outcome measures considered for this review, only three trials reported the duration of mechanical ventilation (Brochard 1998; Brower 1999; Stewart 1998). In the ARDS Network study, this outcome was reported as the median number of days, without further description. However, the absence of a strict protocol for weaning in all the trials, apart from the ARDS Network study (ARDS Network 2000), makes this outcome difficult to evaluate objectively, although important for clinical and economic implications. Similarly, organ failures were reported using different measures. Two trials did not report this outcome at all (Amato 1998; Brower 1999). Brochard 1998 reported that 24 patients in each group suffered from organ failures without further details, whereas Stewart 1998 reported a mean value of two organ failures per patient in each group. ARDS Network 2000 reported the number of days without non-pulmonary organ failure at day 28. Other secondary outcomes considered relevant for this review were not reported in the studies.

Excluded studies

We found two additional studies that were of potential relevance from our updated search (2007 to September 2012) (Meade 2008; Mercat 2008). These studies used low tidal ventilation in both arms. The groups differed by the level of PEEP (Mercat 2008) or the level of PEEP and recruitment manoeuvres (Meade 2008). Therefore, these trials were eventually excluded. Data from one trial (ARDS Network 2000) were used for subsequent multiple publications (see 'Additional Table 1'). We excluded these papers.

Table 1. Secondary publications
Publications
Eisner MD, Thomson T, Hudson LD, Luce JM, HaydenD, Schoenfeld D, et al Efficacy of low tidal volume ventilation in patients with different clinical risk factors for acute lung injury and the acute respiratory distress syndrome. American Journal of Respiratory and Critical Care Medicine 2001;164:231-6.
Ely EW, Wheeler AP, Thompson T, Ancukiewicz M, Steinberg KP, Bernard GR. Recovery rate and prognosis in older persons who developed acute lung injury and the acute respiratory distress syndrome. Annals of Internal Medicine 2002;136:25-36.
Eisner MD, Thompson BT, Schoenfeld D, Anzueto A, Matthay MA; Acute Respiratory Distress Syndrome Network. Airway pressures and early barotrauma in patients with acute lung injury and acute respiratory distress syndrome. American Journal of Respiratory and Critical Care Medicine 2002;165:978-82.
Cheng IW, Eisner MD, Thomson BT, Ware LB, Matthay MA; Acute Respiratory Distress Syndrome Network. Acute effects of tidal volume strategy on hemodynamics, fluid balance, and sedation in acute lung injury. Critical Care Medicine 2005;33:239-40.
Parsons PE, Eisner MD, Thomson BT, Matthay MA, Ancukiewicz M, Bernard GR, et al. Lower tidal volume ventilation and plasma cytokine markers of inflammation in patients with acute lung injury. Critical Care Medicine 2005;33:1-6.

Risk of bias in included studies

All included studies were randomized controlled trials. All the studies found similarity of the two study groups at the time of randomization with respect to important determinants of outcome.

All the studies reached a high level of quality (Figure 2).

Figure 2.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

In five studies (Amato 1998; ARDS Network 2000; Brochard 1998; Brower 1999; Villar 2006) sample size was calculated before the beginning of the study. With the exception of the Amato trial, which reached the target sample (53 patients), the other four studies were stopped early. Brochard 1998 based the calculation on a 30% to 40% reduction of hospital mortality. They estimated a sample of 240 patients and the trial was stopped at 116 because protective ventilation was clearly detrimental (RR = 1.23). Brower 1999 used an estimation of treatment effect based on reversal of respiratory failure as the target outcome for the calculation of sample size. The study was designed to detect a 30% reduction in respiratory failure. They estimated a sample of 130 patients, but the trial was stopped at 52 because the second interim analysis showed that it was unlikely that beneficial treatment effects on reversal of respiratory failure could be demonstrated if the trial was to continue. In ARDS Network 2000, the calculation of sample size was based on a 10% reduction in cumulative mortality at day 180. The estimated sample consisted of 1000 patients and the trial was stopped after enrolling 861 patients because the interim analysis demonstrated that lower tidal volume ventilation was beneficial. The ARDS Network trial, in its design, tried to detect the smallest difference in mortality between groups. Villar 2006 calculated sample size based on a 20% reduction in absolute mortality. The trial was stopped after enrolling 103 patients due to the benefits of low tidal volume ventilation. None of the trials reported a long-term follow up.

Allocation

Five included studies had low risk of selection bias in the random sequence generation domain. One study had an unclear risk (concealment of allocation was unclear) (Brower 1999).

Blinding

In one trial (Amato 1998) it was clearly stated that the study was not blind. No mention of blinding was found in all the other studies. Protocols for management of mechanical ventilation were found in all the studies, thus minimizing performance bias. A protocol for weaning was found only in the ARDS Network study (ARDS Network 2000). For outcomes assessed in this review, it was unclear to whom the blinding referred and so there was an unclear risk of detection bias in all the included studies.

Incomplete outcome data

Most studies included complete follow up and intention-to-treat analysis. In the ARDS Network trial 31 patients (3.6%) were lost to follow up (ARDS Network 2000). Twenty-two of them were patients who were still hospitalized when the trial was stopped, and for nine patients the outcome was unknown (legend, Figure 1 in the original trial). See the 'Characteristics of included studies' for a more detailed description of individual trial quality.

Selective reporting

Risk of reporting bias was rated as low in all trials.

Other potential sources of bias

No other potential sources of bias were found.

Effects of interventions

The ARDS Network trial coordinator and the corresponding author of the Brochard study were requested to supply further data on mortality at day 28. Unfortunately this information has not been made available. Therefore, we extracted the numerical data from figure 1 and figure 5 as presented in the original publications (ARDS Network 2000; Brochard 1998). The total number of randomized patients was 1297. Globally, when considering mortality at the end of the follow-up period, as reported by the trialists, the results from the trials clustered between a RR of 0.6 and 1.23, with overlapping CIs (see Additional Table 2). Overall, the test for heterogeneity yielded a borderline result (P = 0.10; 5 degrees of freedom). The I2 was above the established threshold for combining the data but within the range of 'moderate heterogeneity' (I2 = 45.9%); therefore, random-effects models were applied for calculating an overall estimate (see 'Comparisons and data' table).

Table 2. Results of individual studies
StudyEffect NameTreatedControlTotalEffectLower 95%CIUpper 95%CI
Amato 1998Hospital mortality13/2917/24530.630.391.02
Amato 1998Mortality at day 2811/2917/24530.540.310.91
ARDS Network 2000Mortality at day 180133/432170/4298610.780.650.93
Brochard 1998Mortality at day 6027/5822/581161.230.801.89
Brower 1999Hospital mortality13/2612/26521.080.621.91
Stewart 1998Hospital mortality30/6028/601201.070.741.55
Villar 2006Hospital mortality17/5025/45950.610.380.98

Primary outcomes

Overall mortality at the end of the follow-up period for each trial showed a trend toward a reduction but it did not reach statistical significance. Using a random-effects model, the RR was 0.86 (95% CI 0.69 to 1.06), absolute risk difference of -6% (95% CI -16% to 3%) with moderate heterogeneity (I2 = 45.9%) (see 'Comparison and data 01'). The different duration of follow up used in each trial for outcome assessment, one trial followed patients up to 180 days, one trial had 60 days follow up, one trial used 28 days and the remaining trials used 'in-hospital' mortality, should suggest caution when interpreting this overall estimate of effect. The legend to Figure 1 of the ARDS Network trial stated that 31 patients were censored from the count. Assuming that the 31 patients were approximately evenly distributed, then the actual number of patients reported was: treatment 416 (432 less 16) and control 414 (429 less 15). The calculated deaths were: treatment 129 and control 165. Without further explanation from the authors, this was a conservative approach to getting an estimate of the counts actually used in their analysis. Comparisons 11 and 12 showed that the best-worse analysis did not show any significant effect of the 31 censored patients on the overall results.

We stratified the trials based on comparable outcomes. Using hospital mortality (Amato 1998; ARDS Network 2000; Brower 1999; Stewart 1998; Villar 2006) in 1181 patients, the point estimate was in favour of the low tidal volume ventilation with both the fixed-effect and random-effects models: RR 0.80 (95% CI 0.69 to 0.92), absolute RD -8% (95% CI -13% to -2%); and RR 0.81 (95% CI 0.66 to 0.98), absolute risk reduction -9% (95% CI -18% to 0.0%), respectively. The I2 was 30.8% (see 'Comparison and data 03').

Mortality at day 28 (Amato 1998; ARDS Network 2000; Brochard 1998) in 1030 patients was significantly lower in the patients with the lung protective ventilation strategy, fixed-effect model RR of 0.74 (95% CI 0.61 to 0.88); the absolute risk difference was -10% (95% CI -15% to -4%). When applying a random-effects model the RR of 0.73 was largely unchanged (95% CI 0.61 to 0.87) but the upper limit of the 95% CI of the RD approached unity (absolute RD -12%, 95% CI -23% to 0%). The test for heterogeneity gave a P value of 0.22 (df = 3), I2 = 31.8% and a P value of 0.41 (df = 2), I2 = 0% for hospital mortality and mortality at day 28, respectively, thus confirming more homogeneous data (see 'Comparisons and data 02').

Secondary outcomes

Amongst the secondary outcomes, there were only sufficient data to assess the association between the lung protective ventilation strategy and the duration of mechanical ventilation (see Additional Table 3). Three trials looked at this endpoint (Brochard 1998; Brower 1999; Stewart 1998) and enrolled a total of 288 patients. No additional data were requested from the authors. There was a trend toward a lower duration of mechanical ventilation in patients with the protective ventilation strategy, but this reduction was not statistically significant (fixed-effect model WMD -0.83, 95% CI -1.92 to 0.27, random-effects model WMD 0.38, 95% CI -3.06 to 3.82; test for heterogeneity P = 0.22, df = 3, I2 = 33.1%).

Table 3. Secondary outcomes considered for this review
OutcomesAmato 1998ARDS network 2000Brochard 1998Brower 1999Stewart 1998Villar 2006
Development of MOFN/A15±11 versus 12±11 (days without organ failure)24/58 versus 24/58 (no of patients)N/A2 versus 2 (no of organs/patients)0.3 versus 1.2 (no of organ failure post-pre randomization)
Duration of mechanical ventilationN/A8-10 both groups (median)23±20 versus 21±1611±2.2 versus 12±1.916.6±39 versus 9.7±10N/A
Total duration of mechanical supportN/AN/AN/AN/AN/AN/A
Total duration of stay in intensive careN/AN/AN/AN/A20±39 versus 14±16N/A
Total duration of stay in hospitalN/AN/AN/AN/A33±48 versus 27±26N/A
Long-term mortalityN/AN/AN/AN/AN/AN/A
Long-term health-related quality of lifeN/AN/AN/AN/AN/AN/A
Long-term cognitive outcomesN/AN/AN/AN/AN/AN/A
CostsN/AN/AN/AN/AN/AN/A

Subgroup analyses

We found insufficient data in the trials to perform subgroup analyses assessing the effects of age, severity of disease, and aetiology of ARDS and ALI. Similarly, the studies did not report enough data to perform a subgroup analysis based on concomitant treatments. Subgroup analysis based on the severity of the disease also was not feasible because all the studies included patients with more severe impairment of lung function (PaO2:FiO2 200 mm Hg or less). ARDS Network 2000 included patients with less severe impairment (PaO2:FiO2 between 200 and 300 mm Hg) but these patients accounted for only 15% to 18% of the total sample. Finally, subgroup analysis based on the underlying risk factor was not practicable because each trial included ARDS and ALI from any cause, and in all the studies the authors reported only aggregate data.

We performed subgroup analysis based on the delivery of the interventions and comparing the overall estimate of effect of treatment on all-cause mortality at the end of the follow-up period between the trials with a 'low pressure' control group (mean plateau pressure 31 cm H2O or less) (Brochard 1998; Brower 1999; Stewart 1998) in 288 patients with the effect of treatment in trials with a 'high pressure' control group (plateau pressure greater than 31 cm H2O mean value) (Amato 1998; ARDS Network 2000; Villar 2006) in 1009 patients. Overall mortality was significantly lower in the lung protective ventilation group when a 'higher' plateau pressure in the control arm was applied (fixed-effect model RR 0.74, 95% CI 0.63 to 0.87, absolute risk reduction -11%, 95% CI -17% to -5%) and not significantly different with a 'lower' plateau pressure in the control arm (fixed-effect model RR 1.13, 95% CI 0.88 to 1.45, absolute risk reduction 6%, 95% CI -6% to 17%). The test for heterogeneity yielded I2 = 0% for the 'low' and 'high' plateau pressure groups (see 'Comparisons and data 05').

Publication bias

Although publication bias is not the only cause of an asymmetrical plot, the symmetrical shape of the funnel plot of precision by effect size showed that the overall effect of meta-analysis in this review was not affected by publication bias or biased inclusion criteria. However, the small numbers of trials included should suggest a cautious interpretation.

Sensitivity analysis

We excluded one trial that was clearly unblinded and that used a different method to deliver the intervention in the treatment group (Amato 1998) in 53 patients. The test for heterogeneity changed slightly (P = 0.13, df = 3). Overall mortality at the end of the follow-up periods was also changed slightly (fixed-effect model RR 0.87, 95% CI 0.75 to 1.01, absolute RD -5%, 95% CI -11% to 0%; random-effects model RR 0.97, 95% CI 0.76 to 1.24, absolute RD -2%, 95% CI -11% to 8%). We evaluated the impact of the largest study (ARDS Network 2000) in 861 patients by excluding it from the analysis. The effect of lung protective and conventional treatment strategies on all-cause mortality disappeared (fixed-effect model RR 0.92, 95% CI 0.75 to 1.12; random-effects model RR 0.90, 95% CI 0.67 to 1.15).

The ARDS Network investigators reported that nine patients (hospital mortality unknown) and 22 additional patients (still hospitalized when the trial was stopped) were censored in the survival plot (Legend, Figure 1) (ARDS Network 2000). There was no later information about the eventual outcome of those patients. We performed a sensitivity analysis imputing values for the 31 dropouts according to 'best-worse' case analysis (see 'Comparisons and data' 11 and 12). For the best-case analysis, the 16 and 15 patients were added back and all added control patients were assumed to have died with no additional deaths in the treatment group. Thus the counts were: treatment (129 plus 0)/(416 plus 16) = 129/432; and control (165 plus 15)/(414 plus 15) = 180/429. The test for heterogeneity was P = 0.04, I2 = 59.1%. Therefore, we applied the random-effects model with a RR of 0.89 (95% CI 0.68 to 1.16). For the worst-case analysis, the 16 and 15 patients were added back and all added treatment patients were assumed to have died with no additional deaths in the control group. Thus the counts were: treatment (129 plus 16)/(416 plus 16) = 145/432 and control (165 plus 0)/(414 plus 15) = 165/429. The test for heterogeneity was P- = 0.25, I2 = 25.7%. Therefore, we applied the fixed-effects model: RR 0.92 (95% CI 0.80 to 1.06). In both cases, the CIs reached the no-difference line; the shift of events from one case to the other did not change the statistical significance.

Other secondary outcomes

Although there is evidence that the cause of death in ARDS is the development of organ failure (Monchi 1998), only three studies reported that outcome, each of them using different criteria to define organ failure thus preventing us from combining the results in a summary estimate. In only two studies (ARDS Network 2000; Villar 2006) the ventilatory strategy affected the development of organ failure, reduced in the protective arm, whereas in the 'non-beneficial' studies (Brochard 1998; Stewart 1998) the occurrence of MOF was similar in both groups.

The long-term cognitive outcome and costs were not assessed by the trials.

Discussion

Can we attribute all the differences between the two groups to differences in the tidal volumes used? This is a controversial issue. A study by the ARDS Network (Brower 1999) (the so-called ALVEOLI study) reported no difference in outcome when ARDS patients were ventilated with low tidal volume and high or low PEEP. That means that PEEP is not as important as tidal volume in changing outcomes. Villar 2006 confirms this finding as when a high-low PEEP and large tidal volume difference was used between groups the difference in outcomes appeared again. Therefore, the tidal volume difference clearly had an impact on mortality. Two recent trials (Meade 2008; Mercat 2008) compared ventilation with lower or higher PEEP in patients ventilated with the routine use of low tidal volumes, finding no difference in mortality. This finding confirms that tidal volume is the major determinant of outcome.

Meta-analysis performed by stratifying trials according to 'high plateau pressure' and 'low plateau pressure' in the control group confirmed that when delivery of conventional tidal volume was associated with a plateau pressure of 31 cm H2O or less there was no evidence of decreased mortality from protective ventilation. The large difference in sample size of the two groups (288 versus 1009) calls for cautious interpretation.

Plateau pressure reflects both pulmonary and chest compliance and, therefore, the same limit of pressure may reflect lower tidal volumes if the chest is stiff, and vice versa. Patients with potentially altered chest compliance were excluded in all the studies. Nevertheless, in some patients with ARDS flattening of the pressure-volume (PV) curve may be due to an increase in chest wall elastance related to abdominal distension (Ranieri 1997). Intra-abdominal pressure (a possible confounding factor) was not accounted for or controlled in the selected studies. Although abdominal pressure, a sort of baseline variable, should be distributed evenly between the groups, intra-abdominal pressure should be measured in each ARDS patient to establish whether plateau pressure actually reflects pulmonary pressure rather than a decrease in chest wall compliance.

Summary of main results

In our previous updated review we found evidence that a ventilation strategy using a tidal volume equal or less than 7 ml/kg of measured body weight and plateau pressure less than 31 mm H2O reduced mortality at day 28. In recent trials, ventilation with higher tidal volumes has not been used any more, confirming that the so-called lung protective ventilation strategy gained enough evidence to be considered effective in reducing mortality (Esteban 2008). When meta-analysis was repeated using a homogeneous outcome (mortality at day 28), it showed a relevant, stable benefit from the protective ventilatory approach (relative risk reduction = 26%). The benefit from low tidal volume ventilation was maintained, although reduced, when considering hospital mortality as the endpoint. The effect of lower tidal volumes on the development of organ failure (the main determinant of long-term outcome) is uncertain. Conversely, lower tidal volumes are effective on a short-term endpoint, protecting aerated lung parenchyma, leading to lung recovery. The quantitative summary estimate did not show a statistically significant effect of protective ventilation on the duration of mechanical ventilation. Two studies reporting follow up at one to two years were unable to demonstrate that a limited ventilation strategy improves either long-term function or quality of life in survivors of ARDS and ALI (Cooper 1999; Orme 2003). Health-related quality of life and the long-term cognitive outcome were not considered in the trials selected for this review.

In all the selected studies, although the primary purpose of the investigators was to compare two different tidal volumes other elements of the ventilatory strategy were associated with it. The experimental intervention included permissive hypercapnia, variable levels of PEEP and low plateau airway pressure. The traditional intervention consisted of higher tidal volume, normocapnia, lower levels of PEEP and potentially higher plateau pressures. Some of these co-interventions were unavoidable and consequent to the nature of the main intervention. Therefore, the studies performed a comparison of two approaches rather than two single interventions and caution is required in interpreting these results, especially when analyses have been inspired by looking at the available aggregate data, which makes it difficult to assess the respective importance of each factor.

A possible interpretation of the discordant results, as proposed by Gattinoni (Gattinoni 2002), could involve variations of trans-pulmonary pressure, which is the distending force of the lung, in the individual patient. The high volume might induce lung damage when the resulting trans-pulmonary pressure is high. Conversely, when trans-pulmonary and airway pressure are within the safe limits high or intermediate tidal ventilation (8 to 10 ml/kg) could be used thus avoiding potentially deleterious effects of low tidal volume, Thus, when chest wall compliance is low a higher plateau pressure may be necessary to reach the same trans-pulmonary pressure, without an increase in tidal volume.

Some questions still remain open. The treatment may be effective only in a subgroup of patients. The lower tidal volume ventilation may be clinically worthwhile only in the more severely ill patients.

Quality of the evidence

The issue of adverse effects of a lower tidal volume was not tackled in the studies. Specifically, the impact of acidosis and hypercapnia on the development of organ failure was not clear.

Ventilation with lower tidal volume can be very effective for short-term lung recovery but its impact on the development of organ failure and long-term recovery is still uncertain or unknown.

Potential biases in the review process

The statistic I2 (45.9%) showed that an important percentage of variability in point estimates is due to heterogeneity rather than sampling error. That value is sufficiently high to justify using a random-effects model. Although the point estimate is similar to that given by the fixed-effect model, the random-effects confidence intervals are wider, the treatment effect is non-significant, and this result is more conservative. The discordant figures also reflect an unstable conclusion (Deeks 2001). Several hypotheses can explain and interpret these data. These are as follows.

There is a 'hidden' source of heterogeneity. Although statistical tests of heterogeneity were not significant, we showed that studies were clinically different in some points. Furthermore, most of the trials did not report protocols of concomitant treatments and associated diseases (that is ventilator-associated pneumonia).

There is a subgroup of patients that can be ventilated with lower tidal volumes or volumes in the conventional range, as long as the plateau pressure is kept below 31 cm H2O, without differences in mortality. Because alveolar recruitment occurs during tidal inflation (Gattinoni 1995), a reduction of tidal volume may prove beneficial when it prevents hyperinflation and over-distention but it could be harmful if it is unable to recruit previously collapsed or compressed alveoli. In this situation, individual titration of ventilation is crucial.

Agreements and disagreements with other studies or reviews

A previous meta-analysis (Eichacker 2002) investigated whether differences in treatment effect could be explained by differences in plateau pressure associated with either the control or low tidal volumes. The authors concluded that treatment in controls differed from current practice in terms of too-high tidal volumes and plateau pressure, and this difference may have influenced the outcomes in two trials (Amato 1998; ARDS Network 2000). Although the method applied in the work by Eichacker has been questioned (Petrucci 2003), his study suggests that as long as tidal volumes produce airway pressures considered safe (31 cm H2O or less), there is no benefit from using lower tidal volumes.

Authors' conclusions

Implications for practice

Overall, the relative risk of death at day 28 is reduced by using ventilation with a lower tidal volume. Hospital mortality is also beneficially affected but there is insufficient evidence to draw any conclusions about morbidity and long-term outcomes. Ventilation with a higher tidal volume and higher plateau pressure is associated with increased risk of death, but the independent contribution of higher tidal volume (over-distension) or higher plateau pressure (barotrauma) cannot be identified. Lower tidal volume ventilation may be preferable when lung recovery is a priority.

Implications for research

Further data are required to assess the long-term health-related quality of life, long-term cognitive outcomes and cost. Large, adequately designed trials with subgroup analysis would be able to determine whether an 'intermediate' tidal volume of 8 to 10 ml/kg IBW would be beneficial, too.
Alternatively, a systematic review that uses individual patient data (IPD) can achieve the ultimate aim to:

  • Undertake survival and other time-to-event analyses. If individual survival times were available for each trial, heterogeneity of the considered outcomes (a major issue of this review) could be solved and the issue whether protective ventilation leads to a prolongation of long-term survival or only to lung recovery could be investigated.

  • Undertake subgroup analysis to assess differences in hypercapnia, actually administered tidal volumes, and other confounders.

Acknowledgements

We would like to thank Prof Marcus Müllner, Prof Nathan Pace, Dr Asima Bokhari, Dr Mark Davies, Janet Wale, Nete Villebro and Kathie Godfrey for commenting on the original review (Petrucci 2004a), and Dr Harald Herkner and Prof Nathan Pace for commenting on the updated review. We would like to thank Karen Hovhannisyan for help in searching the literature.

Data and analyses

Download statistical data

Comparison 1. Protective versus conventional
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Mortality at the end of the follow up period for each trial61297Risk Ratio (M-H, Fixed, 95% CI)0.83 [0.72, 0.95]
2 Mortality at day 2831030Risk Ratio (M-H, Fixed, 95% CI)0.74 [0.61, 0.88]
3 Hospital mortality51181Risk Ratio (M-H, Fixed, 95% CI)0.80 [0.69, 0.92]
4 Duration of mechanical ventilation3288Mean Difference (IV, Fixed, 95% CI)-0.83 [-1.92, 0.27]
5 Mortality at different plateau pressure in control groups6 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
5.1 low-pressure control (< or = 31 cm H2O)3288Risk Ratio (M-H, Fixed, 95% CI)1.13 [0.88, 1.45]
5.2 high-pressure control (> 31 cm H2O)31009Risk Ratio (M-H, Fixed, 95% CI)0.74 [0.63, 0.87]
6 Mortality at the end of the follow-up period for each trial - best case61297Risk Ratio (M-H, Fixed, 95% CI)0.79 [0.69, 0.90]
7 Mortality at the end of the follow-up period for each trial - worst case61297Risk Ratio (M-H, Fixed, 95% CI)0.89 [0.78, 1.02]
Analysis 1.1.

Comparison 1 Protective versus conventional, Outcome 1 Mortality at the end of the follow up period for each trial.

Analysis 1.2.

Comparison 1 Protective versus conventional, Outcome 2 Mortality at day 28.

Analysis 1.3.

Comparison 1 Protective versus conventional, Outcome 3 Hospital mortality.

Analysis 1.4.

Comparison 1 Protective versus conventional, Outcome 4 Duration of mechanical ventilation.

Analysis 1.5.

Comparison 1 Protective versus conventional, Outcome 5 Mortality at different plateau pressure in control groups.

Analysis 1.6.

Comparison 1 Protective versus conventional, Outcome 6 Mortality at the end of the follow-up period for each trial - best case.

Analysis 1.7.

Comparison 1 Protective versus conventional, Outcome 7 Mortality at the end of the follow-up period for each trial - worst case.

Appendices

Appendix 1. CENTRAL and The Cochrane Library search strategy

#1 MeSH descriptor Respiratory Distress Syndrome, Adult explode all trees
#2 acute lung injury
#3 Adult Respiratory Distress Syndrome
#4 Acute Respiratory Distress Syndrome
#5 ARDS or ALI
#6 (#1 OR #2 OR #3 OR #4 OR #5)
#7 MeSH descriptor Tidal Volume explode all trees
#8 artificial near ventilation
#9 tidal volume
#10 protective near ventilation
#11pressure-limited
#12 LPVS
#13 (#7 OR #8 OR #9 OR #10 OR #11 OR #12)
#14 (#6 AND #13)

Appendix 2. Search strategy for MEDLINE (OvidSP)

1. exp Respiratory Distress Syndrome, Adult/ or Adult Respiratory Distress Syndrome/ or Acute Lung Injury/ or Acute Respiratory Distress Syndrome/ or ARDS.mp. or ALI.mp.
2. exp tidal volume/ or exp respiration, artificial/ or tidal volume.mp. or (protective adj3 ventilat*).mp. or (pressure* adj3 limited*).mp. or LPVS.mp.
3. 1 and 2
4. ((randomized controlled trial or controlled clinical trial).pt. or randomized.ab. or placebo.ab. or clinical trials as topic.sh. or randomly.ab. or trial.ti.) not (animals not (humans and animals)).sh.
5. 3 and 4

Appendix 3. Search strategy for EMBASE (OvidSP)

1. respiratory distress syndrome/ or respiratory distress/ or acute lung injury/ or adult respiratory distress syndrome/ or acute respiratory failure/ or adult respiratory distress syndrome.mp. or acute lung injury.mp. or acute respiratory distress syndrome.mp. or (ards or ali).mp.
2. tidal volume/ or artificial ventilation/ or tidal volume.mp. or (protective adj3 ventilation).ti,ab. or pressure* limited*.mp. or LPVS.ti,ab.
3. 1 and 2
4. (placebo.sh. or controlled study.ab. or random*.ti,ab. or trial*.ti,ab. or ((singl* or doubl* or trebl* or tripl*) adj3 (blind* or mask*)).ti,ab.) not (animals not (humans and animals)).sh.
5. 3 and 4

Appendix 4. Search strategy for CINAHL (EBSCOhost)

S1 ( (MH "Respiratory Distress Syndrome, Acute") OR (MH "Respiratory Distress Syndrome") OR (MH "Acute Lung Injury") ) OR ( ARDS or ALI ) OR ( (Respiratory Distress Syndrome and (acute or adult)) )
S2 (MM "Tidal Volume") OR TI ( artificial and ventilation ) OR TX tidal volume OR AB ( protective and ventilation ) OR AB pressure limited
S3 S1 and S2

Appendix 5. Search strategy for ISI Web of Science

#1 TS=(adult respiratory distress syndrome or acute lung injury or acute respiratory distress syndrome or ards or ali)
#2 TS=(tidal volume) or TS=(protective SAME ventilation) or TS= LPVS
#3 TS=(random* or placebo* or multicenter* or prospective) or TS=(trail* SAME (clinical or controlled))
#4 #1 and #2 and #3

What's new

DateEventDescription
15 October 2012New search has been performedWe reran our search from 2006 to 2012. No additional trials eligible for inclusion were found. We updated our methods.
15 October 2012New citation required but conclusions have not changedNew author joined review team: Carlo De Feo.

History

Protocol first published: Issue 4, 2002
Review first published: Issue 3, 2003

DateEventDescription
18 February 2009Amendedminor editing of text
2 August 2008AmendedConverted to new review format.
20 April 2007New search has been performedThis review has been updated in the following ways:

1. A new literature search was run covering trials published up to October 2006.
2. One new trial has been published since the previous version of the review.
3. This trial was included in the analysis. The results from this trial confirmed that low tidal volumes improve survival in acute respiratory distress syndrome (ARDS).
4. The title of the review has changed from "Ventilation with lower tidal volumes versus traditional tidal volumes in adults for acute lung injury and acute respiratory distress syndrome" to "Lung protective ventilation strategy for the acute respiratory distress syndrome". This is because it emerged from the updating that additional (but subordinate) factors other than low tidal volume are involved. The favourable outcome depends on low volume AND low pressure. Positive end expiration pressure (PEEP) is not a determinant of outcome. Thus, as it is a strategy rather than a single intervention the title has been changed to a more general one.

Contributions of authors

Conceiving the review: Nicola Petrucci (NP)
Co-ordinating the review: NP
Undertaking manual searches: NP and CDF
Screening search results: NP
Organizing retrieval of papers: NP
Screening retrieved papers against inclusion criteria: NP
Appraising quality of papers: NP and CDF
Abstracting data from papers: NP and CDF
Writing to authors of papers for additional information: NP
Data management for the review: NP
Entering data into Review Manager (RevMan): NP
RevMan statistical data: NP
Double entry of data: (data entered by person one: NP ; data entered by person two: CDF)
Interpretation of data: NP and CDF
Statistical inferences: NP
Writing the review: NP
Guarantor for the review (one author): NP

Declarations of interest

None known

Sources of support

Internal sources

  • Azienda Ospedaliera Desenzano, Desenzano del Garda, Italy.

External sources

  • No sources of support supplied

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Amato 1998

MethodsProspective, multi-centre, controlled, RCT.
Concealment of treatment allocation: yes.
Generation of allocation sequences: sealed envelopes and a 1:1 assignment scheme.
Intent to treat: yes.
Blinding of treatment: no.
Blinding of outcome assessment: not stated.
Participants53 adults > 14 and < 70 years old (2 centres).
Included: LISS ≥ 2.5, PaO2:FiO2 < 200, Pwedge < 16 mmHg.
Exclusion: Previous lung or neuromuscular disease, MV > 1 week, terminal disease, previous barotrauma, previous lung biopsy or resection, intracranial hypertension, uncontrollable and progressive acidosis, documented coronary insufficiency.
Time period of study: December 1990 - June 1995.
InterventionsTidal volume < 6 ml/kg (n=29) or tidal volume 12 ml/kg (n=24).
Plateau pressure < 20 mmHg (experimental arm) or unlimited (control arm).
PEEP 2 cmH2O above LIP or titrated to best PaO2:FiO2.
OutcomesPrimary: mortality at day 28, in-hospital mortality.
Secondary: development of MOF, barotrauma.
NotesBicarbonate if pH < 7.20. Use of PV curve to set Vt and PEEP. PCV. NNT=4.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskAdequate
Allocation concealment (selection bias)Low riskAdequate
Blinding (performance bias and detection bias)
All outcomes
High riskUnblinding stated
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskData not provided to judge
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskData not provided to judge
Incomplete outcome data (attrition bias)
All outcomes
Low risk 

ARDS Network 2000

MethodsProspective, multi-centre, controlled, RCT.
Concealment of treatment allocation: yes.
Generation of allocation sequences: centralized interactive voice system. Each research coordinator had a unique PIN.
Intent to treat: yes.
Double blind: not stated.
Participants861 adults > 18 years old (10 centres).
Included: AECCC, PaO2:FiO2 < 300 mmHg, Pwedge < 18 mmHg.
Exclusion: Pregnancy, high intracranial pressure, sickle cell disease, COPD, weight > 1 kg/cm height, neuromuscular disease, burns over more than 30% BSA, bone marrow or lung transplantation, chronic liver disease.
Time period of study: March 1996 - March 1999.
InterventionsTidal volume 6 mml/kg or 12 ml/kg of IBW.
Plateau pressure ≤ 30 mmHg or ≤ 50 mmHg.
PEEP titrated to SaO2 88-95%.
OutcomesPrimary: mortality before hospital discharge or at day 180.
Secondary: development of MOF, duration of mechanical ventilation, ventilator-free days, barotrauma.
NotesPowered for 1000 patients. Stopped early because of statistically significant benefit in the experimental arm (fourth interim analysis).
Increased ventilation if pH < 7.15. NNT=12.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk 
Allocation concealment (selection bias)Low riskAdequate
Blinding (performance bias and detection bias)
All outcomes
Unclear riskData not provided to judge
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskData not provided to judge
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskData not provided to judge
Incomplete outcome data (attrition bias)
All outcomes
Low risk 

Brochard 1998

MethodsProspective, multi-centre, controlled, RCT.
Concealment of treatment allocation: yes.
Generation of allocation sequences: sealed envelopes.
Intent to treat: yes.
Double blind: not stated.
Participants116 patients ≤ 76 years old (25 centres).
Inclusion: LISS ≥ 2.5, PaO2:FiO2 < 200 mmHg, Pwedge < 18.
Exclusion: History of left heart failure, cardiogenic oedema, organ failure other than the lung, need for epinephrine > 1mg/hr or norepinephrine > 2 mg/hr, COPD or liver failure or renal failure, moribund state, intracranial hypertension, head injury, chest wall abnormalities.
Time period of study: January 1994 - September 1997.
InterventionsTidal volume 7 ml/kg or 10 ml/kg.
Plateau pressure between 25-30 mmHg or peak pressure ≤ 60 mmHg.
PEEP titrated to PaO2:FiO2 > 200 mmHg.
OutcomesPrimary: mortality at day 60.
Secondary: development of MOF, duration of mechanical ventilation, barotrauma.
NotesUse of nitric oxide allowed. Bicarbonate if pH < 7.05. NNH=12.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk 
Allocation concealment (selection bias)Low riskAdequate
Blinding (performance bias and detection bias)
All outcomes
Unclear riskData not provided to judge
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskData not provided to judge
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskData not provided to judge
Incomplete outcome data (attrition bias)
All outcomes
Low risk 

Brower 1999

MethodsProspective, multi-centre, controlled, RCT.
Concealment of treatment allocation: yes.
Generation of allocation sequences: yes, block design, method not stated.
Intent to treat: yes.
Double blind: not stated.
Participants52 patients ≥ 18 years old.
Included: AECCC, PaO2:FiO2 < 200 mmHg, Pwedge < 18 mmHg.
Exclusion: Pregnancy, Acute neurologic disease, life expectancy < 3 months, COPD, sickle cell disease, lobectomy or pneumonectomia.
Time period of study: not stated.
InterventionsTidal volume 8 ml/kg or (10-12) mm/kg of IBW.
Plateau pressure ≤ 30 mmHg or ≤ (45-55) mmHg.
PEEP titrated to SaO2 86-94%.
OutcomesPrimary: in-hospital mortality.
Secondary: duration of mechanical ventilation, reversal of respiratory failure, barotrauma.
NotesBicarbonate if pH < 7.30. NNH=26.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskMethod not stated
Allocation concealment (selection bias)Low riskAdequate
Blinding (performance bias and detection bias)
All outcomes
Unclear riskData not provided to judge
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskData not provided to judge
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskData not provided to judge
Incomplete outcome data (attrition bias)
All outcomes
Low risk 

Stewart 1998

MethodsProspective, multi-centre, controlled, RCT.
Concealment of treatment allocation: yes.
Generation of allocation sequences: block design, computer-generated random numbers tables.
Intent to treat: yes.
Double blind: not stated.
Participants120 patients ≥ 18 years old (8 centres).
Included: High risk for ARDS, PaO2/FiO2 < 250 mmHg.
Exclusion: MV expected to be required for > 48 hrs, PIP > 30 cmH2O for 2 hrs or more before randomization; moribund state; cardiogenic pulmonary oedema myocardial ischaemia; pregnancy; known intracranial abnormalities.
Time period of study: not stated.
InterventionsTidal volume 8 ml/kg or (10-15) ml/kg of IBW.
Peak pressure 30 mmHg or 50 mmHg.
PEEP titrated to 89-93%.
OutcomesPrimary: in-hospital mortality.
Secondary: development of MOF, duration of mechanical ventilation, total duration of stay in intensive care, total duration of stay in hospital, barotrauma.
NotesBicarbonate if pH < 7.00, or increased ventilation if refractory acidosis. NNH=30.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk 
Allocation concealment (selection bias)Low riskAdequate
Blinding (performance bias and detection bias)
All outcomes
Unclear riskData not provided to judge
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskData not provided to judge
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskData not provided to judge
Incomplete outcome data (attrition bias)
All outcomes
Low risk 

Villar 2006

  1. a

    RCT: randomized controlled trial. PaO2: arterial partial pressure of oxygen. FiO2: fraction of inspired oxygen. PaO2:FiO2: hypoxia score. LISS: lung injury scoring system. AECCC: American-European consensus conference committee. IBW: ideal body weight. PEEP: positive end-expiration pressure. MOF: multi-organ failure. PCV: pressure controlled ventilation. NNT: number needed to treat. NNH: number needed to harm. Vt: tidal volume. PIP: peak inspiratory pressure. LIP: low inflection point. PV: pressure-volume.
    In the ARDS Network trial 31 patients were lost to follow up and were censored from the count.

MethodsProspective, multi-centre, controlled, RCT. Concealment of allocation: yes.
Generation of allocation sequences: sealed envelopes.
Intent to treat: yes.
Double blind: not stated.
Participants103 patients > 15 years old (8 centres). Included: AECCC, Included: AECCC, PaO2:FiO2 < 200 mmHg, Pwedge < 18 mmHg.
Exclusion: Pregnancy, severe neurologic damage, cancer patients in terminal stage, high risk of mortality within 3 months.
Time period of study: March 1999 - March 2001.
InterventionsTidal volume (5-8) ml/kg (n=53) or tidal volume (9-11) ml/kg (n=50).
PEEP 2 cmH2O above LIP or > 5 cmH2O, titrated to SaO2 > 90%.
OutcomesPrimary: ICU mortality. Secondary: in-hospital mortality, ventilator-free days, non-pulmonary organ dysfunction, barotrauma.
NotesUse of PV curve to set PEEP.
Stopped early because of statistically significant benefit in the experimental arm.
NNT=5.
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk 
Allocation concealment (selection bias)Low riskAdequate
Blinding (performance bias and detection bias)
All outcomes
Unclear riskData not provided to judge
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskData not provided to judge
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskData not provided to judge
Incomplete outcome data (attrition bias)
All outcomes
Low risk 

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
  1. a

    RCT: randomized controlled trial. ARDS: acute respiratory distress syndrome.

Brower 2004This trial used low tidal volumes in both the study groups. Thus, it is not eligible for inclusion because it did not compare different tidal volumes but only different levels of PEEP.
Esteban 2000This RCT has not been designed to test MV with lower tidal volume in ARDS. Tidal volume and plateau pressure in the experimental arm overlap tidal volume and plateau pressure in the control group.
Meade 2008This RCT has not been designed to test MV with lower tidal volume, as both groups used lower tidal volume. The groups differed by the levels of PEEP and recruitment manoeuvres.
Mercat 2008This RCT has not been designed to test MV with lower tidal volume, as both groups used lower tidal volume. The groups differed only by the levels of PEEP.
Ranieri 1999This RCT tested the hypothesis that protective ventilation minimizes pulmonary and systemic cytokine response and was not specifically designed to assess the benefit on mortality of lower tidal volume ventilation. Although the context of interventions in both groups met the inclusion criteria for this review, the protocol was designed to be a 36-hour study, and 28-day mortality was reported as a post hoc analysis. In addition, analysis was not conducted on an intention-to-treat basis. Loss to follow up was 14%.
Rappaport 1994Prospective, non-blinded RCT. Analysis was not conducted on intention-to-treat basis (attrition bias). Tidal volumes overlap in both groups.

Ancillary