This review assessed the effects of noninvasive positive pressure ventilation (NPPV) for patients with acute hypoxaemic respiratory failure not due to cardiogenic pulmonary oedema. The authors concluded that NPPV appeared to be beneficial but its effects, particularly on mortality, may vary among different patient populations. The review was well conducted and the results are likely to be reliable.

To assess the effect of noninvasive positive pressure ventilation (NPPV) on the rate of endotracheal intubation, length of stay in the intensive care unit (ICU) and hospital, and mortality for patients with acute hypoxaemic respiratory failure not due to cardiogenic pulmonary oedema.

MEDLINE (1980 to October 2003), EMBASE (January 1990 to October 2003), the Cochrane Controlled Trials Register, DARE, the Cochrane Database of Systematic Reviews and the Cochrane Database of Methodology Reviews were searched; the search terms were reported and no language restrictions were applied. In order to identify further studies, the conference abstracts of several relevant meetings (1990 to 2003) were handsearched, the reference lists of identified studies and reviews were checked, and experts were contacted.

Study designs of evaluations included in the review

Only randomised controlled trials (RCTs) were eligible for inclusion.

Specific interventions included in the review

Studies of noninvasive ventilation plus standard therapy versus standard therapy alone were eligible for inclusion. All included studies were of pressure-cycled ventilation using either conventional ventilators or ventilators specifically designed for noninvasive ventilation.

Participants included in the review

Studies were eligible if more than 60% of the participants had acute hypoxaemic respiratory failure not associated with cardiogenic pulmonary oedema or an exacerbation of chronic obstructive pulmonary disease (COPD) and not requiring immediate ventilatory support. In two of the included studies all participants met these criteria, while in two others, data were available separately for participants meeting the criteria. In the other studies, only a small proportion of participants had COPD or cardiogenic pulmonary oedema. Studies were either of immunocompromised patients, postlung resection patients, patients with community-acquired pneumonia or postextubation respiratory failure, or involved a heterogeneous population.

Outcomes assessed in the review

The studies were required to report one or more of the following outcomes: the need for endotracheal intubation; the length of stay in the ICU or hospital; ICU or hospital mortality. All of the included studies reported the need for endotracheal intubation and ICU or hospital mortality; each of the other outcomes was also reported by several of the included studies.

How were decisions on the relevance of primary studies made?

Two independent reviewers selected the studies.

Methodological quality was assessed on 11 criteria, with studies scoring a point for each criterion adequately addressed. The criteria were randomisation, allocation concealment, blinding, patient selection, comparability of the groups at baseline, treatment protocol, confounders, cointerventions, outcome definitions, extent of follow-up and intention-to-treat analysis. Two independent reviewers performed the quality assessment. Any disagreements were resolved by consensus or after consultation with a third reviewer.

Two independent reviewers extracted the data. Any disagreements were resolved by consensus or after consultation with a third reviewer. For mortality outcomes and need for endotracheal intubation, the risk difference and 95% confidence interval (CI) were extracted from each study. For length of stay outcomes, the mean length of stay and 95% CI were extracted.

How were the studies combined?

For each outcome, the studies were pooled in a meta-analysis, using a random-effects model, to give a summary risk difference or weighted mean difference with 95% CIs. Funnel plots were used to assess the likelihood of publication bias.

How were differences between studies investigated?

Heterogeneity was assessed both visually and using the chi-squared test. Sensitivity analyses were performed to assess the impact of removing studies involving COPD or cardiogenic pulmonary oedema patients from the analyses. The potential influence of baseline risk of mortality on treatment effect was investigated visually by plotting control group mortality against the risk difference between the intervention and control groups.

The review included 8 RCTs. Data from 366 participants were included.

The methodological quality of the studies was variable: 3 studies scored 10, 2 scored 8, 2 scored 7 and one scored 4.

Endotracheal intubation (8 RCTs): NPPV was associated with a significantly lower rate of endotracheal intubation, with a risk reduction of 23% (95% CI: 10, 35). The exclusion of 2 studies with COPD or cardiogenic pulmonary oedema patients gave similar results (risk reduction 24%, 95% CI: 8, 36).

Length of ICU stay (7 RCTs): NPPV was associated with a reduction in length of ICU stay of 1.9 days (95% CI: 1.0, 2.9). The exclusion of 2 studies with COPD or cardiogenic pulmonary oedema patients gave the same result.

Length of hospital stay (3 RCTs): NPPV was associated with a significant increase in length of hospital stay of 2.8 days (95% CI: 0.9, 4.7).

ICU mortality (8 RCTs): NPPV was associated with a reduction in ICU mortality of 17% (95% CI: 8, 26). The exclusion of 2 studies with COPD or cardiogenic pulmonary oedema patients gave a similar result (reduction 16%, 95% CI: 5, 27).

Hospital mortality (5 RCTs): NPPV did not have a statistically significant effect on hospital mortality. Statistical heterogeneity (significance level set at P<0.05) was not detected for any of the meta-analyses. However, a visual inspection of the forest plots suggested both qualitative and quantitative differences among the study results.

The results suggested a positive relationship between control group mortality and the effect NPPV has on survival, with a more prominent effect for ICU mortality than for hospital mortality.

Funnel plots did not suggest any evidence of publication bias.

The addition of NPPV to standard therapy for patients with acute hypoxaemic respiratory failure appeared to decrease the need for intubation, ICU length of stay and ICU mortality, but the effects, particularly on mortality, may vary greatly among different patient populations.

The review stated well-defined inclusion criteria in order to address a clear question. A comprehensive search strategy without language restrictions was used to identify the primary studies, meaning it was unlikely that relevant studies were missed. In addition, funnel plots were used to assess the likelihood of publication bias. Two independent reviewers selected the studies and extracted the data, which should have further minimised the introduction of bias and errors. Although some details of the primary studies were reported, the actual results were presented only as forest plots with limited numerical information. Appropriate methods were used to combine the studies and to investigate both statistical and clinical heterogeneity. Overall, the authors' conclusions are likely to be reliable, and their caution in making recommendations appears appropriate.

Practice: The authors stated that the routine use of NPPV for acute respiratory failure was not supported by current evidence from RCTs. They suggested that NPPV should be strongly considered for immunosuppressed or post-thoracotomy patients, but that its use in other patient groups should be carefully monitored to ensure a beneficial effect.

Research: The authors stated that future RCTs should focus on specific patient groups with hypoxaemic respiratory failure rather than more heterogeneous populations.

This additional published commentary may also be of interest. Carroll GC. Review: noninvasive positive-pressure ventilation reduces intubation and length of ICU stay in acute respiratory failure. ACP J Club 2005;143:19.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.