Five RCTS (n=2,447) were included. Sample size ranged from 53 to 983. One study scored the maximum of 5 points on the Jadad scale, two scored 3 and two scored 2.
High PEEP resulted in a significant reduction in hospital mortality (RR 0.89, 95% CI 0.80 to 0.99; five RCTs). There was no evidence of heterogeneity (p=0.91). Differences in PEEP protocols were not associated with differences in mortality rates. There was a strong negative correlation (correlation coefficient -0.89, p<0.05) between predicted mortality rates and the relative risk of hospital mortality associated with use of high PEEP; this suggested that the higher the predicted mortality the greater the mortality reduction associated with use of high PEEP. There was no association between the use of pressure-volume curves and mortality rates. There was no difference in 28-day mortality (three RCTs), intensive care unit-free days (two RCTs), ventilator-free days (four RCTs), organ failure-free days (two RCTs) or barotrauma (five RCTs) between the two treatment groups.
There was evidence of publication bias based on visual inspection of the funnel plot and using the Egger (p=0.05) and the Begg-Mazumdar tests (p=0.02), which suggested under-publication of negative results.