The effect of prone positioning on mortality in patients with acute respiratory distress syndrome: a meta-analysis of randomized controlled trials

Abstract

Introduction: Prone positioning (PP) has been reported to improve the survival of patients with severe acute respiratory distress syndrome (ARDS). However, it is uncertain whether the beneficial effects of PP are associated with positive end-expiratory pressure (PEEP) levels and long durations of PP. In this meta-analysis, we aimed to evaluate whether the effects of PP on mortality could be affected by PEEP level and PP duration and to identify which patients might benefit the most from PP.

Methods: Publications describing randomized controlled trials (RCTs) in which investigators have compared prone and supine ventilation were retrieved by searching the following electronic databases: PubMed/MEDLINE, the Cochrane Library, the Web of Science and Elsevier Science (inception to May 2013). Two investigators independently selected RCTs and assessed their quality. The data extracted from the RCTs were combined in a cumulative meta-analysis and analyzed using methods recommended by the Cochrane Collaboration.

Results: A total of nine RCTs with an aggregate of 2,242 patients were included. All of the studies received scores of up to three points using the methods recommended by Jadad et al. One trial did not conceal allocation. This meta-analysis revealed that, compared with supine positioning, PP decreased the 28- to 30-day mortality of ARDS patients with a ratio of partial pressure of arterial oxygen/fraction of inspired oxygen ≤ 100 mmHg (n = 508, risk ratio (RR) = 0.71, 95 confidence interval (CI) = 0.57 to 0.89; P = 0.003). PP was shown to reduce both 60-day mortality (n = 518, RR = 0.82, 95% CI = 0.68 to 0.99; P = 0.04) and 90-day mortality (n = 516, RR = 0.57, 95% CI = 0.43 to 0.75; P < 0.0001) in ARDS patients ventilated with PEEP ≥ 10 cmH2O. Moreover, PP reduced 28- to 30-day mortality when the PP duration was >12 h/day (n = 1,067, RR = 0.73, 95% CI = 0.54 to 0.99; P = 0.04).

Conclusions: PP reduced mortality among patients with severe ARDS and patients receiving relatively high PEEP levels. Moreover, long-term PP improved the survival of ARDS patients.

Figure 1

Flowchart of the meta-analysis.

Figure 2

Meta-analysis of the effect of prone positioning on 28- to 30-day mortality in acute respiratory distress syndrome patients related to the ratio of partial pressure of arterial oxygen/fraction of inspired oxygen. The evidence gathered in our meta-analysis shows obvious heterogeneity, which was measured using the Mantel–Haenszel (M-H) χ² test (P = 0.004) and the I² heterogeneity test (I² = 56%). A random-effects model was used. The z-test result for overall effects was not statistically significant (P = 0.007). In the ≤300 mmHg ratio of partial pressure of arterial oxygen/fraction of inspired oxygen (P/F) subgroup, the z-test for overall effects was not statistically significant (P = 0.18). In the subgroup of patients with P/F ratios between 100 and 200 mmHg, the z-test result for overall effects was not statistically significant (P = 0.30). In the subgroup of patients with P/F ≤ 100 mmHg, the z-test result for overall effects was statistically significant (P = 0.003). “Weight” is the contribution of each study to the overall risk ratio. CI, Confidence interval; I2, Percentage of total variation across studies from between-study heterogeneity rather than by chance; PP, Prone positioning; SP, Supine positioning.

Figure 3

Meta-analysis of the effect of prone positioning on 28- to 30-day mortality related to positive end-expiratory pressure in acute respiratory distress syndrome patients. The evidence we gathered shows obvious heterogeneity, which we calculated using the Mantel–Haenszel (M-H) χ² test (P = 0.01) and an I² test (I² = 64%). A random-effects model was used. The z-test result for overall effects was not statistically significant (P = 0.18). In the subgroup of patients with positive end-expiratory pressure (PEEP) < 10 cmH₂O, the z-test result for overall effects was not statistically significant (P = 0.61). In the subgroup of patients with PEEP values between 10 and 13 cmH₂O, the z-test result for overall effects was statistically significant (P = 0.09). “Weight” is the contribution of each study to the overall risk ratio. CI, Confidence interval; I², percentage of total variation across studies from between-study heterogeneity rather than chance; PP, Prone positioning; SP, Supine positioning.

Figure 4

Meta-analysis of the effect of prone positioning on 60-day mortality in acute respiratory distress syndrome patients with positive end-expiratory pressure ≥10 cmH₂O. No obvious heterogeneity was found using the Mantel–Haenszel (M-H) χ² test (P = 0.31) and the I² test (I² = 15%). A fixed-effects model was used. The z-test result for overall effects was statistically significant (P = 0.04). “Weight” is the contribution of each study to the overall risk ratio. CI, Confidence interval; I², Percentage of total variation across studies from between-study heterogeneity rather than chance; PP, Prone positioning; SP, Supine positioning.

Figure 5

Meta-analysis of the effect of prone positioning on 90-day mortality in acute respiratory distress syndrome patients related to positive end-expiratory pressure. The evidence we gathered shows obvious heterogeneity based on the results of the Mantel–Haenszel (M-H) χ² test (P = 0.001) and the I² test (I² = 81%). A random-effects model was used. The z-test result for overall effects was statistically significant (P = 0.33) in the subgroup of patients with positive end-expiratory pressure (PEEP) levels ≥10 cmH₂O. In the PEEP <10 cmH₂O subgroup, the z-test result for overall effects was not statistically significant (P = 0.53). In the subgroup of patients with PEEP levels between 10 and 13 cmH₂O, the z-test result for overall effects was statistically significant (P < 0.0001). “Weight” is the contribution of each study to the overall risk ratio. CI, Confidence interval; I², Percentage of total variation across studies from between-study heterogeneity rather than chance; PP, Prone positioning; SP, Supine positioning.

Figure 6

Meta-analysis of the effect of prone positioning on 28- to 30-day mortality in acute respiratory distress syndrome patients related to the duration of prone positioning. The evidence we gathered shows obvious heterogeneity based on the results of the Mantel–Haenszel (M-H) χ² test (P = 0.01) and the I² test (I² = 64%). A random-effects model was used. The z-test result for overall effects was not statistically significant (P = 0.15). In the subgroup of patients with prone positioning (PP) duration ≥12 h/day, the z-test result for overall effects was statistically significant (P = 0.04). In the subgroup of patients with PP duration <12 h/day, the z-test result for overall effects was statistically significant (P = 0.60). “Weight” is the contribution of each study to the overall risk ratio. CI, Confidence interval; I², Percentage of total variation across studies from between-study heterogeneity rather than chance; SP, Supine positioning.